Revert "Warn about non-ASCII characters in text fields of the job form"

This reverts commit 91d049b4df.
Warn about non-ASCII characters in text fields of the job form
2026-08-03 14:34:46 +03:00 · 2026-08-03 14:34:12 +03:00 · 2026-08-03 14:26:40 +03:00 · 2026-08-03 14:26:17 +03:00 · 2026-08-03 04:28:43 +03:00 · 2026-08-03 04:24:52 +03:00
294 changed files with 40449 additions and 3866 deletions
@@ -60,7 +60,7 @@ jobs:
        run: go install -tags 'postgres' github.com/golang-migrate/migrate/v4/cmd/migrate@v4.17.1

      - name: apply migrations
-        run: migrate -path migrations -database "$TEST_DATABASE_URL" up
+        run: migrate -path internal/storage/postgres/migrations -database "$TEST_DATABASE_URL" up

      - name: integration tests
        run: go test -tags=integration ./internal/storage/postgres/ -v
@@ -0,0 +1,146 @@
+name: release
+
+# Builds static coordinator and worker-agent binaries for every major
+# platform on a v* tag push, attaches them (plus SHA-256 checksums) to the
+# GitHub Release, and pushes the coordinator image to GHCR.
+#
+#   git tag v1.0.0 && git push origin v1.0.0
+on:
+  push:
+    tags: ["v*"]
+
+permissions:
+  contents: write
+  packages: write
+
+jobs:
+  binaries:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [linux, darwin, windows]
+        arch: [amd64, arm64]
+    defaults:
+      run:
+        working-directory: coordinator
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-go@v5
+        with:
+          go-version-file: coordinator/go.mod
+          cache-dependency-path: coordinator/go.sum
+
+      - name: vet
+        run: go vet ./...
+
+      - name: build coordinator and worker-agent
+        env:
+          VERSION: ${{ github.ref_name }}
+        run: |
+          mkdir -p dist
+          for cmd in coordinator worker-agent; do
+            CGO_ENABLED=0 GOOS=${{ matrix.os }} GOARCH=${{ matrix.arch }} \
+              go build -trimpath \
+                -ldflags="-s -w -X main.version=${VERSION#v}" \
+                -o "dist/${cmd}-${{ matrix.os }}-${{ matrix.arch }}${{ matrix.os == 'windows' && '.exe' || '' }}" \
+                "./cmd/${cmd}"
+          done
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: binaries-${{ matrix.os }}-${{ matrix.arch }}
+          path: coordinator/dist/*
+          if-no-files-found: error
+
+  wheel:
+    runs-on: ubuntu-latest
+    steps:
+      # Full history: setuptools_scm derives the package version from the
+      # release tag (v1.1.0-alpha.16 -> 1.1.0a16), so the wheel is
+      # version-locked to the binaries of the same release.
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          fetch-tags: true
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: build the scimesh wheel
+        run: |
+          python -m pip install --quiet build
+          python -m build --wheel --outdir dist
+          ls -la dist/
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: wheel
+          path: dist/*.whl
+          if-no-files-found: error
+
+  release:
+    needs: [binaries, wheel]
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/download-artifact@v4
+        with:
+          path: artifacts
+          # Only the binary artifacts: the image job also uploads a buildkit
+          # cache artifact (*.dockerbuild) that download-artifact cannot fetch.
+          pattern: binaries-*
+          merge-multiple: true
+
+      - uses: actions/download-artifact@v4
+        with:
+          name: wheel
+          path: artifacts
+
+      - name: checksums
+        working-directory: artifacts
+        run: sha256sum * > SHA256SUMS.txt
+
+      - uses: softprops/action-gh-release@v2
+        with:
+          files: |
+            artifacts/*
+            install.sh
+            install.ps1
+          # Pre-release tags (e.g. v1.1.0-alpha.1) publish as pre-releases.
+          prerelease: ${{ contains(github.ref_name, '-alpha') }}
+          generate_release_notes: true
+
+  image:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: docker/setup-buildx-action@v3
+
+      - uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - uses: docker/metadata-action@v5
+        id: meta
+        with:
+          images: ghcr.io/${{ github.repository }}/coordinator
+          tags: |
+            type=semver,pattern={{version}}
+            type=semver,pattern={{major}}.{{minor}}
+            type=raw,value=latest
+
+      - uses: docker/build-push-action@v6
+        with:
+          context: coordinator
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          build-args: |
+            VERSION=${{ github.ref_name }}
@@ -16,3 +16,5 @@ test_structures/
 worker-data*/
 scimesh-worker-data/
 coordinator/.demo/
+site/
+coordinator/bin/
@@ -0,0 +1,66 @@
+# Session Goal
+
+выполни план на ночь (полный план ниже — автономное исполнение, пользователь спит)
+
+## Ночная сессия — полный план (пользователь спит, 2026-08-03)
+
+### 1. Баги (реальные, найденные в проде)
+- [ ] **Пустое имя воркера**: регистрация принимает `name=""` (в БД пользователя было 15 таких). Фикс: валидация в `domain.NewWorker` (TrimSpace != "") + регрессионный тест. Частично начат — в `worker_test.go` сломан тест (`fixedTime` vs `testNow`), доделать.
+- [ ] **`worker-agent --check` не видит venv**: проверяет только системный python3; после Install воркер работает через `~/.scimesh-worker/venv`. Выровнять с визардом (пробовать venv, если он есть).
+
+### 2. Визуальный долг — рестайлинг старых страниц в дизайн-систему админки (#0b0e13, карточки, pill-статусы, кнопки)
+- [ ] `new-job.html` (форма запуска вычислений, UIElement-поля сохранить)
+- [ ] `job.html` (детали джоба: шарды, артефакты, прогресс, JS-логику сохранить)
+- [ ] `workloads.html`
+- [ ] `add-worker.html`
+- [ ] `profile.html`
+- [ ] Браузерная проверка каждой страницы (playwright).
+
+### 3. Технический долг (паритет движков, обещанный планом)
+- [ ] **Postgres integration-тесты для admin-методов**: `SetTrust`, `WorkloadSettings` (Get/List/Set), `ListJobsPaginated`, метрики (`JobCountsByDay`, `TaskStats`, `ArtifactSizeByKind`, `DatabaseSizeBytes`) — сейчас покрыт только sqlite. (build tag integration, docker postgres как в CI.)
+
+### 4. Фичи из плана (v2-задел)
+- [ ] **Статистика воркера в визарде**: статусная страница — claimed/completed/failed/heartbeat, парсинг из `worker.log` (обещано в docs/ui-admin-worker-plan.md §4).
+- [ ] **Admin: prune артефактов** (danger zone): удаление артефактов completed-джобов старше N дней (строки + blob-файлы), освобождает место; кнопка в Settings с подтверждением.
+- [ ] **Admin: удаление offline-воркеров** (устаревшие, как 15 мусорных) — кнопкой в Workers вместо SQL.
+
+### 5. Гигиена
+- [ ] **`pyproject.toml` версия 0.1.0 → setuptools_scm** (версия из git-тегов), убрать sed-костыль из release.yml; локальный `pip install -e .` покажет правильную версию.
+- [ ] Синхронизация STATUS.md / README с фактическим состоянием (админка, визард, wheel, авто-открытие, restyle).
+
+### 6. Верификация и релиз
+- [ ] Полный гейт: `go test -race ./...`, golangci-lint `--build-tags=integration`, pytest.
+- [ ] Браузерные проверки: админка (все секции), визард (install+start), рестайленные страницы.
+- [ ] E2E quorum-флоу (untrusted воркер через worker key → джоб требует quorum) в Docker, если останется время.
+- [ ] Релиз `v1.1.0-alpha.16` (бинарники + wheel), проверка ассетов.
+- [ ] Обновить этот файл: прогресс по пунктам, COMPLETED в конце.
+
+## Plan (предыдущая задача — выполнена)
+1–7. Docker E2E пайплайна «install как человек → serve → визард → воркер → джоб» — выполнено, см. Progress ниже.
+
+## Progress (ночная сессия)
+- ✅ **П.1 Пустое имя воркера**: `domain.NewWorker` нормализует/отклоняет пустое имя + `TestNewWorkerRejectsBlankName` (починен `fixedTime`→`testNow`).
+- ✅ **П.2 `--check`**: пробует managed venv (если установлен) + реальный пробинг учётки (exchange ключа / claim-пробa) — `CheckAuth` + тесты; на машине пользователя: `✓ auth: credential accepted`, venv python, scimesh installed.
+- ✅ **П.3 Рестайлинг**: единый CSS-partial `ui-base.html` (дизайн-система админки), все 5 страниц (new-job, job, workloads, add-worker, profile) переведены, проверены в браузере без console-ошибок.
+- ✅ **П.4 Postgres integration**: admin-методы (SetTrust, ListJobsPaginated, TaskCounts, byDay/byWorkload, TaskStats, ArtifactSize, DB size, WorkloadSettings) + `ensureMigrated` для порядка запуска; весь suite зелёный.
+- ✅ **П.5 Статистика воркера в визарде**: лог `task claimed` в агента + парсинг registered/claimed/completed/failed → `/api/status.stats` + карточки в статусной странице + тест.
+- ✅ **П.6 Prune артефактов**: `JobRepository.ListCompletedBefore/Delete` (sqlite+postgres+memstore), usecase `PruneArtifacts` (каскад + blob-файлы), `POST /ui/admin/api/prune`, кнопка в Settings, тесты (sqlite+usecase); E2E: 200, freed bytes.
+- ✅ **П.7 Удаление offline-воркеров**: `WorkerRepository.Delete` + `Admin.RemoveWorker` (только offline) + `POST /ui/admin/api/workers/{id}/remove` + кнопка в Workers + тест; E2E: 204, строка удалена.
+- ✅ **П.8 setuptools_scm**: `dynamic = ["version"]`, CI-джоба wheel без sed (fetch-depth 0); локальная проверка: wheel на теге = `scimesh-1.1.0a16-py3-none-any.whl` (совпадает с Go-нормализацией); релизный ассет подтверждён.
+- ✅ **П.9 Docs**: STATUS.md синхронизирован (админка, визард, wheel, CTX-19/20 implemented).
+- ✅ **П.10 (доп.) Баг в serve-режиме**: worker-key exchange был недоступен снаружи (userservice на loopback) — добавлен прокси `POST /worker-tokens/exchange` на координаторе, `PublicUserserviceURL=""` + fallback на origin в add-worker. Проверено E2E.
+- ✅ **П.10 Quorum E2E (Docker)**: 2 untrusted-воркера с разными ключами (alice/bob) → джоб completed 3/3, в task_results по 2 голоса от разных владельцев с одинаковым sha256 → результат байт-в-байт = локальному эталону.
+- ✅ **Финальный гейт**: `go test -race ./...` ✅, golangci-lint 0 issues ✅, pytest 208 ✅, postgres integration ✅, Windows кросс-сборка ✅.
+- ✅ **Релиз v1.1.0-alpha.16** (бинарники + wheel `scimesh-1.1.0a16`), все воркфлоу success; бинарники на машине пользователя обновлены до alpha.16.
+
+## Progress (прошлая работа — выполнено)
+- ✅ Релизы alpha.12–15: фикс версии визарда, venv task_runner, preflight через venv, MkdirAll при скачивании wheel, кнопка Install в шаблоне.
+- ✅ Docker E2E: координатор+воркер контейнеры, установка install.sh, визард (config→install→start), воркер online, джоб completed, результат байт-в-байт = локальному эталону.
+- ✅ На машине пользователя: визард alpha.15, правильный токен, venv из wheel, воркер emil-pc online, 15 пустых воркеров вычищены из БД.
+- ✅ Гейт: race + lint + pytest 208.
+
+## Completion
+COMPLETED — ночной план выполнен полностью (10 пунктов + 2 найденных бага, включая E2E quorum на релизном коде). Все гейты зелёные, релиз v1.1.0-alpha.16 опубликован.
+
+## Completion (предыдущая задача)
+COMPLETED — пайплайн доведён до рабочего состояния и проверен на релизных артефактах v1.1.0-alpha.14.
@@ -5,8 +5,10 @@
 SciMesh is a Python package for molecular-similarity workloads. Source lives in
 `scimesh/`: `chemistry/` reads data and makes fingerprints, `workloads/`
 contains commands, and `core/` provides the workload protocol and registry.
-The worker daemon in `scimesh/worker/` is a coordinator client, not a database
-client. Tests are in `tests/`; specifications in `docs/`; roadmap: `PLAN.md`.
+The Go worker agent (`coordinator/internal/agent/`) is a coordinator client,
+not a database client; the Python side of a claimed task lives in
+`scimesh/worker/` (the per-task SDK execution entry). Tests are in `tests/`;
+specifications in `docs/`; roadmap: `PLAN.md`.

 For distributed work, read `.agents/`, `docs/api-contract.md`,
 and `STATUS.md`. Use one CTX task per pull request; local workloads are the
@@ -1,19 +1,32 @@
 .DEFAULT_GOAL := help

-.PHONY: help demo-ui demo-down demo-logs
+.PHONY: help agent coordinator demo-ui demo-down demo-logs smoke-two-worker docs docs-serve

 help:
 	@printf '%s\n' \
 	  'SciMesh developer commands:' \
-	  '  make demo-ui                 Start the local UI pipeline demo with 2 workers.' \
-	  '  make demo-ui WORKERS=3       Start the demo with 3 local workers.' \
+	  '  make agent                   Build the Go worker agent (coordinator/bin/worker-agent).' \
+	  '  make coordinator             Build the coordinator server as a static binary' \
+	  '                              (coordinator/bin/coordinator).' \
+	  '  make demo-ui                 Start the local UI pipeline demo with 2 Go worker agents.' \
+	  '  make demo-ui WORKERS=3       Start the demo with 3 workers.' \
 	  '  make demo-logs               Follow coordinator logs for the demo.' \
 	  '  make demo-down               Stop demo containers and workers.' \
+	  '  make smoke-two-worker        E2E: two Go agents process 4 shards and the' \
+	  '                              result must match the local CLI reference.' \
+	  '  make docs                    Build the MkDocs site into site/.' \
+	  '  make docs-serve              Serve the MkDocs site at http://localhost:8000.' \
 	  '' \
 	  'After make demo-ui: open http://localhost:18080/ui (operator / demo-ui-secret).'

 # Convenient entry points from the repository root. Extra settings are passed
 # through, for example: make demo-ui WORKERS=3
+agent:
+	$(MAKE) -C coordinator agent
+
+coordinator:
+	$(MAKE) -C coordinator coordinator
+
 demo-ui:
 	$(MAKE) -C coordinator demo-ui

@@ -22,3 +35,12 @@ demo-down:

 demo-logs:
 	$(MAKE) -C coordinator demo-logs
+
+smoke-two-worker:
+	./scripts/two-worker-smoke.sh
+
+docs:
+	.venv/bin/mkdocs build
+
+docs-serve:
+	.venv/bin/mkdocs serve
@@ -5,11 +5,13 @@
 > platform. It is intentionally detailed enough to split into independent task
 > briefs for developers or coding agents.
 >
-> **Planning baseline.** This branch starts from `Workers`: the Python package
-> has local `similarity-search` and `similarity-graph` workloads plus a Worker
-> Daemon client. The coordinator and PostgreSQL implementation do not yet
-> exist. The Worker contract and the Go/PostgreSQL design briefs in `docs/` are
-> part of this plan.
+> **Current planning baseline (2026-08-01).** The Go/PostgreSQL coordinator,
+> Python Worker Agent, versioned distributed-workload protocol, artifact-backed
+> task lifecycle, reducer orchestration, operator UI, User Service, and
+> distributed `similarity-search` are implemented on `main`. The evidence-based
+> completion tracker is [`STATUS.md`](STATUS.md); this document defines the
+> remaining direction and dependencies. Earlier descriptions of a missing
+> coordinator are historical context, not current work.

 ---

@@ -51,7 +53,7 @@ Coordinator reducer -> final artifact -> download/status API

 ### 2.1 In scope

- Go 1.22+ coordinator service with PostgreSQL 15+;
+- Go 1.25+ coordinator service with PostgreSQL 15+;
 - Python Worker Daemon running existing SciMesh workloads locally;
 - durable job, task, worker, and artifact metadata;
 - local coordinator-managed artifact storage for the first deployment;
@@ -66,7 +68,8 @@ Coordinator reducer -> final artifact -> download/status API

 - cloud object storage, Kubernetes, autoscaling, and multi-region operation;
 - arbitrary shell commands sent by coordinator to workers;
- user accounts, multi-tenancy, billing, or sophisticated authorization;
+- billing and sophisticated multi-tenant administration beyond the implemented
+  User Service and owner scoping;
 - GPU scheduling and multiprocessing inside a worker;
 - Docker as a required runtime dependency;
 - video/CV processing implementation;
@@ -490,14 +493,84 @@ brute-force graph for both `greater` and `less` threshold directions.

 ### 7.3 Future workload policy

-A new workload is accepted only when it supplies:
+A workload is more than a runner. It must define validation, planner and
+reducer behavior, worker allowlist/capabilities, input and output artifacts,
+UI/API parameters, reproducible execution environment, result verification,
+golden cross-worker fixtures, and applicable resource limits. The future public
+contract is described in [`docs/scimesh-sdk-roadmap.md`](docs/scimesh-sdk-roadmap.md).
+Its normative future interfaces and execution semantics are defined in the
+design-draft [`docs/scimesh-sdk-contract.md`](docs/scimesh-sdk-contract.md).

- an input/parameter validator;
- an explicit sharding strategy;
- bounded-memory task execution;
- deterministic reduction semantics;
- fixture-based local and distributed correctness tests;
- a `describe()` payload for UI/API discovery.
+Every workload declaration must classify its task decomposition and input/output
+artifact shapes, determinism, reduction semantics, verifier mode, supported
+trust modes, CPU/memory/accelerator needs, and maximum output growth. The
+initial profiles are:
+
+| Profile | Current acceptance policy |
+| --- | --- |
+| Byte-exact deterministic | Supported for untrusted quorum when whole artifacts have identical SHA-256. |
+| Canonical-exact deterministic | Deferred until the parser, schema, ordering, encoding, and serializer are versioned. |
+| Numeric deterministic with tolerance | Deferred until structured numeric comparison exists. |
+| Stochastic/search-based | Requires domain-specific evidence, repeated runs, or trusted execution. |
+| Trusted-only or domain-verified | May be planned only with an explicit trust policy and verifier. |
+
+The current untrusted quorum records one vote per owner and accepts a task only
+when distinct owners upload artifacts with the same complete-file SHA-256. It
+therefore supports only the byte-exact profile (or a workload that first makes
+its output byte-identical through a specified canonicalization step). Reducers
+must fail safely rather than silently merge inconsistent partial results.
+
+Before a workload is admitted to untrusted execution it needs a reproducibility
+gate: pinned environment/container digest and dependency versions; fixed locale,
+timezone, UTF-8/newline/CSV settings; explicit invalid-row and algorithm
+options; canonical representation and ordering; deterministic filenames/archive
+metadata; golden fixtures from two independently provisioned workers; local vs
+distributed parity; and retry/out-of-order completion tests. A loose dependency
+constraint is insufficient for byte-exact quorum.
+
+Near-term critical path:
+
+```text
+distributed similarity-graph
+  -> reliability, security, and cross-language CI
+    -> stable first release
+      -> SDK foundation and descriptor-batch
+        -> additional deterministic workloads
+```
+
+Initial deterministic-workload backlog: `descriptor-batch` (the first SDK
+reference workload), molecule standardization, SMARTS screening, fingerprint
+export, fixed-template SMIRKS enumeration with strict caps, and reaction
+validation/descriptors. `similarity-graph` remains ahead of this backlog.
+Bounded combinatorial libraries and seeded conformers need specialized controls.
+ML, retrosynthesis, docking, QM, molecular dynamics, and GPU workloads are
+deferred until verifier/trust and reproducibility requirements are met.
+
+### 7.4 Future verification, concurrency, and accelerators
+
+Verification is a future versioned workload capability, not permanent
+whole-file-SHA logic. Planned modes are `ExactArtifactVerifier`,
+`CanonicalRecordVerifier`, `NumericToleranceVerifier`, `DomainSpecificVerifier`,
+and `TrustedWorkerPolicy`. Exact SHA-256 remains the first and safest mode;
+canonical and numeric modes must compare bounded structured data and publish
+sanitized evidence and failure reasons.
+
+Worker concurrency remains **1** until implemented and tested. Its target model
+is one physical machine running one Worker Agent with `N` execution slots and
+one isolated subprocess per active Task, rather than one registered worker per
+CPU core. `max_concurrency` must be separate from `cpu_count`; each task keeps
+its own heartbeat, attempt directory, lease lifecycle, resource request, and
+graceful-drain behavior. CPU-bound scientific code should use processes and
+avoid nested oversubscription.
+
+Accelerator support is also deferred. The coordinator matches generic resource
+requirements; the Worker Agent discovers and isolates devices (including
+`CUDA_VISIBLE_DEVICES`) and owns process/accounting lifecycle; the Python
+workload owns batching, memory strategy, deterministic output, and scientific
+validation; reducers/verifiers define CPU/GPU-independent semantics. CUDA and
+scientific kernels do not belong in the Go coordinator. GPU work follows stable
+CPU slices, generic resource requirements, pinned worker images, and tested
+CPU/GPU or domain-valid equivalence.

 ---

@@ -820,6 +893,212 @@ CTX-09 enables final result downloads.
 - failure/retry scenarios have automated coverage;
 - README contains architecture diagram, security caveat, and troubleshooting.

+### CTX-13 — In-worker CPU parallelism
+
+**Goal:** Allow a worker to use a bounded, configured number of CPU threads or
+processes while preserving the existing one-task-per-lease coordinator model.
+
+**Depends on:** CTX-12.
+
+**Acceptance criteria:**
+
+- worker concurrency is an explicit configuration value with a safe default of
+  one;
+- a task's internal parallel execution has bounded memory and does not build a
+  dense N×N similarity matrix;
+- CPU-parallel `similarity-search` and `similarity-graph` outputs match the
+  single-threaded local reference byte-for-byte where ordering is observable;
+- result ordering is deterministic across worker counts and block sizes;
+- cancellation, lease loss, and worker failure stop child work safely and do
+  not report a successful result;
+- benchmarks and tests cover one-worker and multi-worker configurations.
+
+### CTX-14 — GPU-accelerated workload execution
+
+**Goal:** Add an optional GPU execution backend for supported molecular
+workloads, while retaining the validated CPU implementation as the reference
+and fallback.
+
+**Depends on:** CTX-13.
+
+**Acceptance criteria:**
+
+- GPU capability and backend version are advertised explicitly by a worker;
+- the coordinator schedules GPU work only to compatible workers and CPU-only
+  workers continue to claim CPU tasks;
+- unsupported hardware, unavailable drivers, and GPU execution errors produce
+  sanitized failures or a documented CPU fallback;
+- GPU results match the CPU reference within a documented, tested numerical
+  tolerance and preserve deterministic output ordering;
+- GPU memory use is bounded and no dense N×N similarity matrix is created;
+- CPU-only CI verifies backend selection and contract behavior, with GPU
+  integration tests documented for compatible runners.
+
+### CTX-15 — User Service and access control
+
+**Goal:** Introduce a dedicated User Service for user identity and access
+control, without coupling workers to user credentials or moving scientific
+workload logic into the service.
+
+**Depends on:** CTX-12.
+
+**Acceptance criteria:**
+
+- the service has a versioned, documented API in
+  [`docs/user-service-api-contract.md`](docs/user-service-api-contract.md) and
+  owns user identity data;
+- credentials and authentication tokens are stored and handled securely; they
+  are never logged or exposed to workers;
+- authenticated identity is propagated to coordinator requests through an
+  explicit, validated boundary;
+- authorization restricts access to jobs and artifacts to the intended user or
+  project;
+- unauthenticated, expired-token, and cross-user access attempts have
+  automated failure tests;
+- the existing single-operator demo remains usable through a documented local
+  development configuration.
+
+### CTX-16 — Workload SDK foundation
+
+**Goal:** Provide a strict Python authoring SDK for installed, allowlisted
+scientific workloads while retaining the CTX-07 distributed protocol as a
+compatible wire profile.
+
+**Depends on:** CTX-07 and CTX-08. Coordinator-backed generalized scheduling
+also depends on CTX-10 through CTX-14, but the Python contract and local
+conformance runtime can land independently and must fail closed for unavailable
+features.
+
+**Acceptance criteria:**
+
+- public manifest, workflow, task, artifact, resource, execution, provenance,
+  and verifier value objects are immutable, typed, JSON-safe, versioned, and
+  strict about unknown fields;
+- installed workload discovery requires an administrator allowlist plus exact
+  workload version and package digest; job parameters cannot select code;
+- compatibility negotiation covers SDK/protocol/profile/feature/environment
+  versions and occurs before planner invocation;
+- plans/tasks pin package and manifest digests plus selected trust mode, and
+  quorum candidates carry coordinator-owned candidate/owner and scientific
+  binding identities;
+- `core-batch-v1` has a trusted local conformance executor with atomic resource
+  reservation, sealed-output/provenance validation, declared verifier
+  invocation, and golden scientific parity;
+- exact, canonical-record, and structured numeric-tolerance verifier
+  primitives return bounded sanitized decisions;
+- the existing distributed `similarity-search` is available through an adapter
+  without changing its wire schema, worker alias boundary, or scientific
+  result, and parity is tested;
+- advanced dynamic, stream, accelerator, gang, and side-effect profiles are
+  rejected unless an enforcing runtime advertises their required features;
+- an author guide documents package entry points, security boundaries,
+  conformance tests, and current coordinator/Worker limitations.
+
+---
+
+### CTX-17 — Self-provisioning coordinator and setup wizard
+
+**Goal:** A downloaded coordinator binary should bring up a working platform
+with as little manual configuration as possible: it provisions its own
+schema, and a `setup` command walks the operator through the remaining
+environment (database creation, secrets, admin account).
+
+**Depends on:** the Go coordinator and the release build pipeline. All steps
+are implemented: embedded migrations with `AUTO_MIGRATE`, the `setup` wizard,
+the embedded SQLite storage backend (`SCIMESH_DB=sqlite`), the embedded
+userservice, and the `coordinator serve` single-binary mode with local worker
+agents. The standalone `users/` service and the PostgreSQL engine remain for
+cluster deployments.
+
+**Acceptance criteria:**
+
+- the binary embeds the migrations and applies pending ones on startup by
+  default (`AUTO_MIGRATE=false` opts out for managed databases); applying is
+  idempotent and safe under concurrent starts;
+- `coordinator setup` (interactive, then non-interactive with `--yes`) checks
+  database reachability, offers to create the role/database when credentials
+  allow it, applies migrations, writes a `.env` with a generated `JWT_SECRET`
+  and storage path, and prints exact next steps;
+- `coordinator --version` and the setup output agree on the release build;
+- the wizard explains what it cannot do itself: running PostgreSQL and the
+  userservice, with concrete commands (docker compose, systemd) to finish;
+- the Docker image keeps working without the separate migrate step, and the
+  release workflow publishes the binaries that support `setup`;
+- setup fails closed on non-interactive input and never logs secrets.
+
+---
+
+### CTX-18 — Single-binary platform (`coordinator serve`)
+
+**Goal:** A scientist installs one file, runs one command, and gets the whole
+platform: coordinator, both databases, the userservice, and local workers —
+no PostgreSQL, no Docker, no Python setup.
+
+**Depends on:** CTX-17 (embedded migrations, SQLite, embedded userservice).
+
+**Acceptance criteria:**
+
+- `coordinator serve` provisions `~/.scimesh` (databases, secrets chmod 0600,
+  generated admin password printed once, artifacts dir) and serves the UI on
+  127.0.0.1:8080 by default; `--open` opens the browser;
+- `--workers N` spawns N `coordinator agent` subprocesses that claim and
+  execute tasks locally; agents are stopped on shutdown;
+- the embedded userservice listens on the loopback interface and shares the
+  JWT secret with the coordinator, so UI login/registration work unchanged;
+- a managed scientific runtime venv (`~/.scimesh/venv`) is bootstrapped on
+  first start; `SCIMESH_PIP_PACKAGE` controls what gets installed (the PyPI
+  name is not ours), and `TASK_RUNNER` points at the venv python;
+- `install.sh` / `install.ps1` detect the platform, download the release
+  binary, and print the start command; both are release assets;
+- the PostgreSQL engine and the standalone userservice stay fully supported;
+- `coordinator serve` passes the full local E2E without any external service:
+  health, login, upload, claim, compute, reduction, byte-exact result.
+
+---
+
+### CTX-19 — Coordinator Admin UI
+
+**Goal:** Replace the demo-level operator pages with a real admin console
+(`/ui/admin`): system overview, job management with filters and pagination,
+worker management (trust, capabilities), users/roles and worker keys,
+workload enable/disable, settings (read-only), and storage metrics.
+
+**Depends on:** CTX-11 (UI patterns, session/roles), CTX-15 (userservice
+proxy), the sqlite/postgres storage pair.
+
+**Status: implemented.** Admin console at `/ui/admin`: system/storage/health,
+jobs (filter + pagination + owner resolution), workers with trust controls,
+users & worker keys (userservice admin endpoints), workload enable/disable
+(`workload_settings` migration in both engines, enforced at submit time and
+hidden from the job form), metrics (7-day buckets, failure rate), and
+settings with an audited worker-token reveal. All `/ui/admin/*` routes are
+admin-only and backed by bounded read models; sqlite + postgres parity;
+unit/permission/integration tests green.
+
+### CTX-20 — Worker Setup UI
+
+**Goal:** A local setup wizard embedded in the **worker-agent** binary
+(`worker-agent setup`, browser on 127.0.0.1) that turns any machine into a
+worker: coordinator URL + auth (serve token or worker key), work dir,
+connection check, start/stop, and a live status page. Runs on machines that
+have only the worker installed — no coordinator needed.
+
+**Depends on:** the agent daemon; the worker-key exchange (CTX-15).
+
+**Status: implemented.** `worker-agent setup` serves the local wizard on
+127.0.0.1 (default port 12700): coordinator URL + token or worker key,
+work dir and name, preflight check (coordinator/health, python3, scimesh),
+config saved to `~/.scimesh-worker/config.json` (0600), start/stop of the
+worker as a background process, and a live status page with the agent log.
+The daemon also gained `--config <path>` (environment still wins) and
+`--check`. End-to-end verified: the wizard started a real worker that
+registered with a coordinator and appeared in the admin console.
+
+**Acceptance criteria:** full plan in
+[`docs/ui-admin-worker-plan.md`](docs/ui-admin-worker-plan.md) (section 4);
+the agent gains `setup`, `--config <path>`, and `--check`; end-to-end: the
+wizard starts a real worker that registers and completes a job.
+
 ---

 ## 10. Suggested assignment bundles
@@ -835,6 +1114,7 @@ parallel unless one engineer owns integration.
 | Distributed computation | CTX-07, CTX-08, CTX-10 | Scientific Python engineer |
 | Product surface | CTX-09, CTX-11 | Full-stack/backend engineer |
 | Quality gate | CTX-12 | DevOps/QA engineer |
+| Workload SDK | CTX-16 | Scientific Python/platform engineer |

 Suggested order for a small team:

@@ -948,15 +1228,23 @@ Before merging a task, reviewer checks:
 Do not start these before CTX-12 is accepted.

 - Replace local artifact storage with S3/MinIO behind an `ArtifactStore` API.
- Add worker labels/capacity-aware scheduling and concurrency > 1.
+- Add worker labels and capacity-aware scheduling.
+- Implement CTX-13 for bounded in-worker CPU parallelism.
+- Implement CTX-14 for optional GPU-accelerated workload execution.
+- Implement CTX-15 for the User Service and authenticated user/project access.
 - Add cancellation propagation to workers.
 - Add image outputs and final PDF reporting to job artifacts.
 - Add CV/video workloads using the same planner/runner/reducer contract.
 - Add observability export (Prometheus/OpenTelemetry).
- Add per-user/project authorization and signed artifact URLs.
+- Add signed artifact URLs.
 - Add shard caching and content-addressed input deduplication.
 - Add job priority and fair scheduling.
 - Add a CLI for submitting and monitoring remote jobs.
+- Publish the scimesh Python package to PyPI so the managed venv bootstrap
+  (`SCIMESH_PIP_PACKAGE`) works out of the box on a scientist's machine.
+- Bundle a Python runtime (python-build-standalone) into the release so local
+  workers need no system Python at all.
+- Native installers (.msi/.dmg/.deb) built by the release workflow.

 ---

@@ -1,14 +1,26 @@
 # SciMesh

-SciMesh is a scientific-workload framework for molecular datasets. Its public CLI
-runs exact similarity search and sparse similarity-graph construction locally in
-one Python process; it creates no dense similarity matrix. The Go/PostgreSQL
-coordinator and Python worker can run a shard-based `similarity-search`
-pipeline locally. After every shard succeeds, the coordinator deterministically
-merges its candidates into one final global top-k CSV. See
-[`STATUS.md`](STATUS.md).
+SciMesh is a local-first platform for scientific computation on molecular
+datasets. It turns a scientific run into independent tasks, dispatches them
+to worker agents, and deterministically combines the partial results into a
+checksum-protected final artifact.

-The ChEMBL TSV database is intentionally not included in this repository. Download it separately and pass its path to the commands below. The expected columns are `chembl_id` and `canonical_smiles`.
+- **The Workload SDK (`scimesh.sdk`)** — a strict Python framework for
+  authoring scientific workloads: `similarity-search` (exact top-k Tanimoto),
+  `similarity-graph` (exact sparse graph), `descriptor-batch`, and
+  `molwt-filter`. Workloads are ordinary user scripts built on the SDK; they
+  run locally, in the conformance harness, and on claimed coordinator tasks
+  without touching any other part of the program.
+- **The coordinator and worker agents** — a Go/PostgreSQL coordinator with an
+  operator UI and Go worker agents that execute SDK workloads in a Python
+  subprocess. The UI is workload-agnostic: the "New computation" form offers
+  every workload from the embedded SDK library, and each workload declares its
+  own form controls (`UIElement`) through the SDK.
+
+The ChEMBL TSV database is intentionally not included in this repository.
+Download it separately and pass its path to the commands below. The expected
+columns are `chembl_id` and `canonical_smiles`. See
+[`STATUS.md`](STATUS.md) and [`PLAN.md`](PLAN.md).

 ## Installation

@@ -27,6 +39,66 @@ conda install -c conda-forge rdkit
 pip install -e .
 ```

+## Releases
+
+Every `v*` tag pushes a GitHub Release with static binaries for `coordinator`
+and `worker-agent` on linux/darwin/windows × amd64/arm64 (plus SHA-256
+checksums), the installer scripts, and the `coordinator` image on GHCR:
+
+```bash
+docker pull ghcr.io/emil28092005/SciMesh/coordinator:latest
+```
+
+For scientists: one command downloads the right binary, starts it, and opens
+the UI in the browser:
+
+```bash
+# Linux / macOS — installs, starts and opens the admin console automatically
+curl -fsSL https://raw.githubusercontent.com/emil28092005/SciMesh/main/install.sh | bash
+
+# Windows (PowerShell)
+powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/emil28092005/SciMesh/main/install.ps1 | iex"
+```
+
+Set `SCIMESH_AUTO_START=0` to install without starting anything. The old demo
+control room was removed: `/ui` is the admin console. A standalone
+worker is installed the same way (`bash -s worker`, or
+`SCIMESH_COMPONENT=worker` on Windows); its installer opens the local setup
+wizard (`worker-agent setup`) in the browser automatically.
+
+`coordinator serve` is the single-binary mode: it embeds SQLite (coordinator +
+userservice databases), the userservice itself, and local worker agents
+(`--workers N`, default 1). On first start it generates secrets and the admin
+password under `~/.scimesh`, prints the login, and opens the UI. No
+PostgreSQL, no Docker, no environment variables. The scientific runtime is a
+managed venv (`~/.scimesh/venv`); point `SCIMESH_PIP_PACKAGE` at your scimesh
+wheel to install it automatically.
+
+The coordinator's UI is the **admin console** at `/ui/admin` — cluster health
+and storage, paginated job table, worker fleet with trust controls, users and
+worker keys, workload enable/disable, metrics and the worker token. The job
+form (`/ui/jobs/new`), job detail pages and the workload library complete the
+operator surface; `/ui` redirects to the console, `serve --open` lands on it,
+and login returns you to the page you asked for. The **worker** binary (`worker-agent`) carries its own local setup
+wizard for machines that run only a worker: `worker-agent setup` opens a
+browser wizard at `127.0.0.1` that collects the coordinator URL and
+credential, runs a preflight check, saves `~/.scimesh-worker/config.json` and
+starts/stops the worker with a live log (see the
+[standalone docs](mkdocs/standalone.md)).
+
+Manual download and run of a release binary:
+
+```bash
+curl -L -o coordinator https://github.com/emil28092005/SciMesh/releases/latest/download/coordinator-linux-amd64
+chmod +x coordinator
+./coordinator --version
+```
+
+Cluster deployments keep the PostgreSQL engine (`SCIMESH_DB=postgres` with
+`DATABASE_URL`, or `coordinator setup` to provision it) and the standalone
+userservice (`users/`). `coordinator agent` runs a worker agent from the same
+binary.
+
 ## Quick start

 Run the built-in help command for copy-paste examples of both workloads:
@@ -46,9 +118,9 @@ scimesh similarity-graph --help

 ## Manual pipeline demo

-To inspect the coordinator, Web UI, and distributed `similarity-search`
-pipeline by hand, install development dependencies once and start the isolated
-demo from the repository root:
+To inspect the coordinator, Web UI, and distributed pipeline by hand, install
+development dependencies once and start the isolated demo from the repository
+root:

 ```bash
 python3 -m venv .venv
@@ -56,12 +128,26 @@ python3 -m venv .venv
 make demo-ui
 ```

-Open `http://localhost:18080/ui` and sign in with username `operator` and
-password `demo-ui-secret`. The command starts PostgreSQL, the coordinator, and
-two local reference workers. Upload a small ChEMBL TSV, then use the job page
-to follow shard progress, inspect bounded **Preview CSV** results, and see a
-live processing-speed chart in shards per minute. To change the worker count,
-run `make demo-ui WORKERS=3`; stop everything with `make demo-down`.
+The MkDocs documentation site is served inside the UI at `/ui/docs/`
+(`make docs` builds it from `mkdocs/`; the demo mounts `site/`
+automatically, or set `SCIMESH_DOCS_DIR` for a manual coordinator). The site
+covers the complete Workload SDK: guides (`mkdocs/sdk/`), the full
+auto-generated API reference for `scimesh.sdk` (`mkdocs/api/`), and the
+documentation rules the site is written by (`mkdocs/approach.md`).
+
+Open `http://localhost:18080/ui` and sign in with username
+`root@scimesh.local` and password `rootpassword`. The command starts
+PostgreSQL, the coordinator, and two Go worker agents (built by `make agent`;
+each executes the SDK workload in a Python subprocess). The **New computation**
+form offers every upload-ready workload from the installed library — the
+controls come from each workload's own SDK declarations. Upload a small ChEMBL
+TSV, then use the job page to follow shard progress, inspect bounded
+**Preview CSV** results, and see a live processing-speed chart in shards per
+minute. The **Workloads** page shows the installed SDK workload library
+(descriptions, parameters, and artifact schemas) from the embedded catalog;
+regenerate it with `make workloads-export` (or `scimesh workload export`)
+whenever workloads change. To change the worker count, run
+`make demo-ui WORKERS=3`; stop everything with `make demo-down`.

 Run `make help` to display these commands in the terminal.

@@ -133,9 +219,63 @@ pytest

 The package separates common dataset parsing and fingerprints from independent workloads. Add future workloads through the workload registry without changing the main CLI.

+The coordinator and worker agent are Go modules under `coordinator/` and `users/`:
+
+```bash
+cd coordinator && make coordinator agent && go test ./...
+```
+
+On headless servers (no desktop environment), RDKit needs a few X11
+libraries that desktops already ship: `sudo apt-get install -y libxrender1
+libxext6 libxcursor1 libxfixes3 libxi6 libxrandr2`.
+
+`make check` runs the full gate: vet, lint, race tests, the PostgreSQL
+integration suite, and the two-worker end-to-end smoke script.
+
+## Workload SDK
+
+`scimesh.sdk` is the framework only: strict and immutable workload manifests,
+typed artifact ports, static map/reduce plans, resource eligibility and local
+reservations, exact/canonical/numeric verifier primitives, installed-package
+allowlisting, and a local conformance executor. It contains no scientific
+workload code. Workloads are user scripts built on the SDK: the built-in
+`similarity-search`, `similarity-graph`, `descriptor-batch`, and
+`molwt-filter` live in `scimesh/workloads/` (each a small package with
+`core.py` + `definition.py`), composed by `scimesh/workloads/library.py` and
+registered through `scimesh.workloads` entry points. The Worker Agent executes
+those SDK-built workloads directly (see `scimesh/worker/runners.py`), so the
+same scientific handlers run locally, in conformance, and on claimed
+coordinator tasks. `scimesh workload list` and `scimesh workload run` run any
+SDK workload from the command line; `scimesh workload export` writes the
+coordinator's embedded workload catalog, and `scimesh workload allowlist`
+prints the JSON for `SCIMESH_WORKLOAD_ALLOWLIST`.
+
+Workloads can also declare how they should appear in the coordinator UI:
+a tuple of `UIElement`s (`scimesh.sdk.UIElement`) shapes the "New computation"
+form — widget, label, help, defaults, and ordering — plus the coordinator-side
+reduction mode (`reduction`: `top-k` or `ordered-concat`) and whether a single
+uploaded dataset can drive the workload (`upload_ready`). The strict parameter
+schema stays the authoritative validation contract.
+
+See the
+[SDK author guide](docs/workload-sdk.md), [contract](docs/scimesh-sdk-contract.md),
+and [delivery roadmap](docs/scimesh-sdk-roadmap.md).
+
+Dynamic workflows, real Worker concurrency, coordinator-backed GPU allocation,
+streaming, and gang execution remain fail-closed until their versioned runtime
+features are implemented; declaring those profiles does not silently enable
+them.
+
+The included `LocalCoreBatchExecutor` is a trusted, single-threaded in-process
+conformance harness. It validates scientific parity, sealed outputs, provenance,
+and limits, but intentionally refuses profiles that claim network/process
+isolation, secrets, accelerators, gangs, checkpoints, or retries; those require
+the future enforcing Agent runtime.
+
 ## Team

 - [Emil](https://github.com/emil28092005) — Project Lead
 - [Kristina](https://github.com/kristtma) — Tech Lead
+- [Veniamin](https://t.me/Veniamin_Kt) — Scientific Lead
 - [Arkhip](https://github.com/hIpa-ussr) — Programmer
- [Makar](https://github.com/RERAN4K) — Programmer
+- [Reranchik](https://github.com/RERAN4K) — Programmer
@@ -1,7 +1,8 @@
 # SciMesh Status

-**Updated:** 2026-07-24
-**Branch baseline:** `main` at `6e67daa` (distributed similarity-search)
+**Updated:** 2026-08-03
+**Branch baseline:** `main`; this revision adds the admin console, the worker
+setup wizard, and release-shipped Python wheels.

 ## Current state

@@ -17,11 +18,51 @@ the reference behaviour for future distributed execution:

 The Go coordinator and its PostgreSQL-backed task lifecycle are implemented:
 registration, atomic claiming, lease renewal, artifact storage, dataset
-chunking, result/failure reporting, and job progress. The Python worker now
-uses the live coordinator contract. Completed similarity-search shard results
-are reduced once into a checksum-protected final CSV, which is downloadable
-through the coordinator. The full Go checks (including a fresh migration and
-real PostgreSQL smoke test) passed on 2026-07-24.
+chunking, result/failure reporting, and job progress. The Go worker agent now
+uses the live coordinator contract. Completed shard results are reduced once
+into a checksum-protected final CSV, downloadable through the coordinator.
+
+**Single-binary platform (`coordinator serve`)**: the coordinator now ships
+an embedded SQLite storage backend (`SCIMESH_DB=sqlite`), an embedded
+userservice, and `serve`/`agent` subcommands, so one downloaded binary runs
+the whole platform — coordinator, both databases, UI logins, and local
+workers — with no PostgreSQL, no Docker, and no environment variables. The
+first start provisions `~/.scimesh` (secrets, admin password printed once,
+managed scientific-runtime venv) and opens the UI. `install.sh` / `install.ps1`
+download the right release binary in one command and are release assets. The
+PostgreSQL engine, the `setup` wizard, and the standalone `users/` service
+remain fully supported for cluster deployments. The full E2E passes with zero
+external services: health, UI login, job upload, local agent compute,
+reduction, and a byte-exact final CSV.
+
+The User Service is merged into `main`. It owns user accounts, authentication,
+roles, and verified-contributor status; the coordinator scopes user jobs and
+worker operations to the authenticated owner. Its documented v1 contract is in
+[`docs/user-service-api-contract.md`](docs/user-service-api-contract.md).
+Users can create and revoke worker keys for self-service Worker Agent
+enrollment. Untrusted workers require quorum agreement from distinct owners on
+the complete result-artifact SHA-256 before a task is accepted.
+
+**Admin console** (`/ui/admin`, admin role only): system/storage/health,
+paginated jobs with owner resolution, worker fleet with trust controls and
+offline-worker removal, users and worker keys, workload enable/disable
+(persisted in both engines, enforced at submit time), 7-day metrics, and
+settings with an audited worker-token reveal and artifact pruning for old
+finished jobs.
+
+**Worker setup wizard** (`worker-agent setup`, loopback `127.0.0.1`): browser
+wizard that configures the coordinator URL and credential, runs a preflight
+(coordinator reachability, real credential probe, python, scimesh), installs
+the version-locked `scimesh` wheel from the release into a managed venv
+(`~/.scimesh-worker/venv`), pins the venv task runner, and starts/stops the
+worker with live log and parsed claim/completed/failed counters. The release
+pipeline now also ships the Python wheel (`scimesh-<ver>-py3-none-any.whl`,
+version derived from the git tag via setuptools_scm), and both `coordinator
+serve` and the wizard install it automatically.
+
+The full pipeline was verified end-to-end in containers on release artifacts:
+installer → serve → wizard → registration → real similarity-search job →
+byte-exact result vs the local reference.

 ## Milestone tracker

@@ -29,17 +70,27 @@ real PostgreSQL smoke test) passed on 2026-07-24.
 | --- | --- | --- |
 | CTX-00 API and error contract | Implemented | Contract, OpenAPI, and request examples are in `docs/`. |
 | CTX-01 Go coordinator bootstrap | Implemented | Go service and Docker runtime in `coordinator/`. |
-| CTX-02 PostgreSQL migrations | Implemented | Applied by the Compose migration service. |
-| CTX-03 Transactional queue | Implemented | Real-PostgreSQL integration tests cover atomic claims and concurrency. |
+| CTX-02 PostgreSQL migrations | Implemented | Embedded into the binary (`AUTO_MIGRATE`); the CLI path is still available for managed databases. |
+| CTX-03 Transactional queue | Implemented | Real-PostgreSQL integration tests cover atomic claims and concurrency; the SQLite backend mirrors the semantics. |
 | CTX-04 Worker registry and HTTP API | Implemented | Registration, claim, heartbeat, result, failure, and status endpoints. |
 | CTX-05 Artifact storage | Implemented | Coordinator-owned inputs/results, checksum verification, and upload flow. |
-| CTX-06 Python Worker live-contract alignment | Implemented | Worker completed a real uploaded shard via HTTP on 2026-07-23. |
-| CTX-07 Distributed workload protocol | Implemented | Versioned Python contract models, registry, strict plan validation, and deterministic reduction ordering are in `scimesh/distributed/`. |
-| CTX-08 Distributed similarity-search | Implemented | Python planner resolves `query_id` once, creates deterministic shard plans, worker adapter emits exact partial top-k CSVs/metrics, and reducer matches the local reference. |
-| CTX-09 Reducer and final-result API | Implemented | Atomic `reducing` claim, deterministic coordinator-side top-k reducer, sanitized reducer failure, final artifact persistence, `result_uri`, and final CSV download. |
-| CTX-10 Distributed similarity-graph | Not started | Local reference exists. |
-| CTX-11 Dashboard/operator view | Implemented | Protected live control room: recent-run/worker overview, real pipeline-stage visualization, shard attempts and safe failures, validated similarity-search upload, coordinator artifacts, final-result download, and bounded polling. |
-| CTX-12 Reliability, security, CI | In progress | Unit, race, PostgreSQL integration, and smoke checks exist; CI hardening remains. |
+| CTX-06 Python Worker live-contract alignment | Superseded | The Python worker daemon was removed; the Go worker agent (`coordinator/internal/agent/` + `cmd/worker-agent`, or `coordinator agent`) implements the lifecycle and executes SDK workloads via `scimesh/worker/task.py`. |
+| CTX-07 Distributed workload protocol | Implemented | Versioned Python contract models, registry, strict plan validation, and deterministic reduction ordering. |
+| CTX-08 Distributed similarity-search | Implemented | Planner/worker/reducer match the local reference byte-exactly. |
+| CTX-09 Reducer and final-result API | Implemented | Atomic `reducing` claim, deterministic coordinator-side reducers (`top-k` and `ordered-concat`), sanitized failure, final artifact, `result_uri`. |
+| CTX-10 Distributed similarity-graph | Not started | Local reference exists; the SDK-built local graph workload enforces the pair-coverage invariant. |
+| CTX-11 Dashboard/operator view | Implemented | The demo control room was removed; `/ui` lands on the admin console. Job form/detail, workload library, add-machine and profile pages share the admin design system; MkDocs at `/ui/docs/`. |
+| CTX-19 Coordinator Admin UI | Implemented | `/ui/admin` console (see above), bounded read models, admin-only routes, sqlite+postgres parity with integration tests. |
+| CTX-20 Worker Setup UI | Implemented | `worker-agent setup` wizard with runtime installer, `--config`, `--check` (credential + venv probing), live status. |
+| CTX-12 Reliability, security, CI | In progress | vet, gofmt, race tests, golangci-lint (0 issues), PostgreSQL integration, and smoke checks exist. |
+| CTX-15 User Service and access control | Implemented | User/owner scoping, verified contributors, worker keys, quorum; also embeddable (`coordinator serve`). |
+| CTX-16 Workload SDK foundation | Implemented | Strict immutable manifests/plans/artifacts, digest/trust-pinned tasks, negotiation, verifier primitives, conformance harness. |
+| CTX-17 Self-provisioning + setup wizard | Implemented | Embedded migrations, `coordinator setup`, SQLite backend, embedded userservice, `serve` mode. |
+| CTX-18 Single-binary platform | Implemented | `coordinator serve` (data dir, secrets, admin bootstrap, local agents, managed venv) + `install.sh`/`install.ps1`; full no-external-service E2E green. |
+| SDK roadmap step 3: `descriptor-batch` | Implemented | Byte-identical local/distributed output, quorum verifier test. |
+| SDK-built `similarity-search` and `similarity-graph` | Implemented | SDK-built packages, byte-identical to single-process references. |
+| SDK-built `molwt-filter` | Implemented | Minimal authoring example; also the single-binary E2E workload. |
+| SDK authoring scaffold | Implemented | `MapReduceWorkload` with `UIElement` declarations, `reduction`, `upload_ready`; generic `scimesh workload list|run|export|allowlist` CLI. |

 ## Next recommended assignment

@@ -48,14 +99,27 @@ block-pair planning and reduction for `similarity-graph`.

 ## Known constraints

- The worker/coordinator flow currently accepts both underscore API workload
-  names and hyphenated CLI names while the contract is consolidated.
+- The worker/coordinator flow accepts both underscore API workload names and
+  hyphenated names at the runner boundary; the runner normalizes them.
+- The worker executes SDK-built workloads through `scimesh/worker/runners.py`
+  (a workload-generic v1-wire bridge over `TaskSpec`/`LocalTaskContext`);
+  `query_id` resolution and parameter validation live in the workload itself.
+  `max_rows` is a plan-time option and is rejected per task by the stage
+  projection.
 - A real-stack worker test uses a small `query_smiles` shard. The Python
  planner resolves `query_id` once and shares `query_smiles`; the upload UI
  currently accepts `query_smiles` only.
 - The coordinator accepts uploaded distributed jobs only for
  `similarity-search` with `query_smiles`. It rejects `similarity-graph` until
  CTX-10 supplies cross-shard pair planning.
+- The SDK can execute `core-batch-v1` locally, but the protocol-v1 coordinator
+  still has flat single-input/single-result tasks and no package/resource
+  leases. General DAG, concurrent-Agent, GPU, stream, and gang execution needs
+  a versioned coordinator/Worker rollout; unsupported features fail before
+  planner invocation.
+- The local SDK executor is intentionally trusted and in-process. It does not
+  enforce process/network/timeout/credential isolation and rejects declarations
+  that would require those guarantees.

 ## Update rule

@@ -16,6 +16,9 @@ RUN --mount=type=cache,target=/go/pkg/mod go mod download

 COPY . .

+# Release builds inject the tag via build-arg; local builds stay "dev".
+ARG VERSION=dev
+
 # The cache mounts persist the module cache and the compiler's build cache
 # *across* builds, so a rebuild after a code edit recompiles only what changed
 # instead of the whole dependency tree.
@@ -25,7 +28,7 @@ COPY . .
 RUN --mount=type=cache,target=/go/pkg/mod \
    --mount=type=cache,target=/root/.cache/go-build \
    CGO_ENABLED=0 GOOS=linux go build \
-        -trimpath -ldflags="-s -w" \
+        -trimpath -ldflags="-s -w -X main.version=${VERSION#v}" \
        -o /out/coordinator ./cmd/coordinator

 # --- runtime stage --------------------------------------------------------
@@ -1,6 +1,6 @@
 .DEFAULT_GOAL := help

-.PHONY: help build run test test-integration vet lint tidy check migrate-up migrate-down up down down-clean logs ps rebuild psql smoke demo-ui demo-down demo-reset demo-logs
+.PHONY: help build run test test-integration vet lint tidy check migrate-up migrate-down up down down-clean logs ps rebuild psql smoke agent coordinator setup serve workloads-export demo-ui demo-down demo-reset demo-logs

 # `check` deliberately uses its own Compose project and host ports. This keeps
 # it from connecting to or replacing a developer's local PostgreSQL instance.
@@ -26,6 +26,45 @@ DEMO_WORKERS ?= 2
 WORKERS ?= $(DEMO_WORKERS)
 DEMO_DIR ?= .demo

+# workloads.json is the UI workload catalog, generated from the Python SDK
+# workload library. It is checked in so the binary embeds it; regenerate it
+# whenever workloads or their manifests change (requires the Python venv).
+WORKLOADS_JSON := internal/workloads/workloads.json
+
+# Version injected into the binaries via -ldflags; falls back to "dev".
+VERSION ?= $(shell git describe --tags --always --dirty 2>/dev/null || echo dev)
+LDFLAGS := -s -w -X main.version=$(VERSION)
+
+# The Go worker agent: a static coordinator client that executes SDK
+# workloads in a Python subprocess per claimed task.
+agent:
+	CGO_ENABLED=0 go build -trimpath -ldflags="$(LDFLAGS)" -o bin/worker-agent ./cmd/worker-agent
+	@printf '%s\n' 	  'Built bin/worker-agent. Configure via environment:' 	  '  COORDINATOR_URL, WORKER_AUTH_TOKEN, WORK_DIR, CPU_COUNT, MEMORY_MB,' 	  '  POLL_INTERVAL, REQUEST_TIMEOUT, HEARTBEAT_INTERVAL, CAPABILITIES,' 	  '  TASK_RUNNER, MAX_TASKS, EXIT_WHEN_IDLE, WORKER_NAME, WORKER_ID'
+
+# The coordinator server as a static binary, the same way the Docker image
+# builds it (CGO_ENABLED=0, trimmed). Requires PostgreSQL at runtime.
+coordinator:
+	CGO_ENABLED=0 go build -trimpath -ldflags="$(LDFLAGS)" -o bin/coordinator ./cmd/coordinator
+	@printf '%s\n' \
+	  'Built bin/coordinator. Configure via environment:' \
+	  '  DATABASE_URL, COORDINATOR_ADDR, COORDINATOR_TOKEN, UI_AUTH_TOKEN,' \
+	  '  COORDINATOR_STORAGE_DIR, SCIMESH_DOCS_DIR, JWT_SECRET, USERSERVICE_URL' \
+	  '  (embedded schema migrations run on startup; AUTO_MIGRATE=false disables)'
+
+# Interactive wizard: checks the database, creates it when missing (via
+# POSTGRES_ADMIN_URL or --admin-db), applies the embedded schema, generates a
+# JWT_SECRET, and writes a .env file. Non-interactive: SETUP_ARGS=--yes.
+setup: coordinator
+	./bin/coordinator setup $(SETUP_ARGS)
+
+# The single-binary mode: everything embedded (sqlite + userservice + local
+# workers), no PostgreSQL or Docker. SETUP_ARGS=--workers 2 --open.
+serve: coordinator
+	./bin/coordinator serve $(SETUP_ARGS)
+
+workloads-export:
+	cd .. && .venv/bin/scimesh workload export -o coordinator/$(WORKLOADS_JSON)
+
 help:
 	@printf '%s\n' \
 	  'SciMesh coordinator commands:' \
@@ -34,6 +73,8 @@ help:
 	  '  make demo-logs               Follow coordinator logs for the UI demo.' \
 	  '  make demo-down               Stop the demo services and workers.' \
 	  '  make demo-reset              Stop the demo and wipe its data volumes.' \
+	  '  make workloads-export        Regenerate the embedded UI workload catalog.' \
+  '  make setup                   Interactive one-shot provisioning wizard.' \
 	  '  make test / make vet         Run Go verification.' \
 	  '' \
 	  'Demo UI: http://localhost:18080/ui (login page; admin root@scimesh.local / rootpassword).'
@@ -127,10 +168,10 @@ tidy:
 # DATABASE_URL must be set, e.g.:
 #   export DATABASE_URL='postgres://scimesh:scimesh@localhost:5432/scimesh?sslmode=disable'
 migrate-up:
-	migrate -path migrations -database "$(DATABASE_URL)" up
+	migrate -path internal/storage/postgres/migrations -database "$(DATABASE_URL)" up

 migrate-down:
-	migrate -path migrations -database "$(DATABASE_URL)" down 1
+	migrate -path internal/storage/postgres/migrations -database "$(DATABASE_URL)" down 1

 # --- docker --------------------------------------------------------------
 # `up` starts Postgres, applies migrations, then launches the coordinator.
@@ -0,0 +1,85 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log/slog"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent"
+)
+
+// runAgent implements `coordinator agent`: the worker agent as a subcommand of
+// the same binary, so one file can serve the whole platform. `serve` spawns
+// these for its local workers.
+func runAgent(args []string) error {
+	flags := flag.NewFlagSet("agent", flag.ContinueOnError)
+	flags.Usage = func() {
+		_, _ = fmt.Fprintf(flags.Output(), "usage: coordinator agent [options]\n")
+		_, _ = fmt.Fprintf(flags.Output(), "Runs as a worker agent: claims tasks, executes SDK workloads in a\n")
+		_, _ = fmt.Fprintf(flags.Output(), "Python subprocess, uploads results.\n\n")
+		flags.PrintDefaults()
+	}
+	var (
+		coordinatorURL = flags.String("coordinator-url", os.Getenv("COORDINATOR_URL"), "coordinator base URL")
+		token          = flags.String("token", os.Getenv("WORKER_AUTH_TOKEN"), "worker bearer token")
+		workDir        = flags.String("work-dir", os.Getenv("WORK_DIR"), "worker work directory")
+		name           = flags.String("name", os.Getenv("WORKER_NAME"), "worker name (default: hostname)")
+		workerID       = flags.String("worker-id", os.Getenv("WORKER_ID"), "persistent worker id (optional)")
+		cpuCount       = flags.Int("cpu", envInt("CPU_COUNT", 1), "advertised CPU cores")
+		memoryMB       = flags.Int("memory-mb", envInt("MEMORY_MB", 1024), "advertised memory in MiB")
+		poll           = flags.Duration("poll-interval", 2*time.Second, "claim poll interval")
+		taskRunner     = flags.String("task-runner", os.Getenv("TASK_RUNNER"), "python command + args that run scimesh.worker.task")
+		maxTasks       = flags.Int("max-tasks", envInt("MAX_TASKS", 0), "stop after N completed tasks (0 = unlimited)")
+		exitWhenIdle   = flags.Bool("exit-when-idle", false, "exit when the queue is empty")
+	)
+	if err := flags.Parse(args); err != nil {
+		return err
+	}
+	if flags.NArg() > 0 {
+		return fmt.Errorf("agent takes no positional arguments")
+	}
+	if *coordinatorURL == "" || *token == "" || *workDir == "" {
+		return fmt.Errorf("--coordinator-url, --token, and --work-dir are required")
+	}
+	if *taskRunner == "" {
+		*taskRunner = "python -m scimesh.worker.task"
+	}
+
+	logger := slog.New(slog.NewTextHandler(os.Stderr, nil))
+	config := agent.Config{
+		CoordinatorURL: strings.TrimRight(*coordinatorURL, "/"),
+		Capabilities:   agent.DefaultCapabilities(),
+		Token:          *token,
+		WorkerName:     *name,
+		WorkerID:       *workerID,
+		WorkDir:        *workDir,
+		CPUCount:       *cpuCount,
+		MemoryMB:       *memoryMB,
+		PollInterval:   *poll,
+		RequestTimeout: 30 * time.Second,
+		Heartbeat:      15 * time.Second,
+		TaskRunner:     strings.Fields(*taskRunner),
+		MaxTasks:       *maxTasks,
+		ExitWhenIdle:   *exitWhenIdle,
+	}
+	tokens := agent.NewTokenProvider("", "", config.Token, config.RequestTimeout)
+	client := agent.NewClient(config.CoordinatorURL, tokens, config.RequestTimeout)
+	runner := agent.NewTaskRunner(config.TaskRunner)
+	daemon := agent.NewDaemon(&config, client, runner, logger)
+	return daemon.RunForever()
+}
+
+func envInt(name string, fallback int) int {
+	raw := os.Getenv(name)
+	if raw == "" {
+		return fallback
+	}
+	var n int
+	if _, err := fmt.Sscanf(raw, "%d", &n); err != nil {
+		return fallback
+	}
+	return n
+}
@@ -2,6 +2,8 @@ package main

 import (
 	"context"
+	"flag"
+	"fmt"
 	"log/slog"
 	"os"
 	"os/signal"
@@ -12,11 +14,52 @@ import (
 	"github.com/emil28092005/SciMesh/coordinator/internal/metrics"
 	"github.com/emil28092005/SciMesh/coordinator/internal/storage/blob"
 	"github.com/emil28092005/SciMesh/coordinator/internal/storage/postgres"
+	"github.com/emil28092005/SciMesh/coordinator/internal/storage/sqlite"
 	httptransport "github.com/emil28092005/SciMesh/coordinator/internal/transport/http"
 	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+	"github.com/emil28092005/SciMesh/coordinator/internal/workloads"
 )

+// version is injected at build time (-ldflags "-X main.version=...") and
+// reported by --version. "dev" marks a local build.
+var version = "dev"
+
 func main() {
+	args := os.Args[1:]
+	if len(args) > 0 {
+		switch args[0] {
+		case "setup":
+			if err := runSetup(args[1:]); err != nil {
+				fmt.Fprintln(os.Stderr, "setup:", err)
+				os.Exit(1)
+			}
+			return
+		case "serve":
+			if err := runServe(args[1:]); err != nil {
+				fmt.Fprintln(os.Stderr, "serve:", err)
+				os.Exit(1)
+			}
+			return
+		case "agent":
+			if err := runAgent(args[1:]); err != nil {
+				fmt.Fprintln(os.Stderr, "agent:", err)
+				os.Exit(1)
+			}
+			return
+		case "token":
+			if err := runToken(args[1:]); err != nil {
+				fmt.Fprintln(os.Stderr, "token:", err)
+				os.Exit(1)
+			}
+			return
+		}
+	}
+	showVersion := flag.Bool("version", false, "print the build version and exit")
+	flag.Parse()
+	if *showVersion {
+		fmt.Println("coordinator " + version)
+		return
+	}
 	// All work happens in run() so its defers (pool.Close, log flush, signal
 	// stop) still execute: os.Exit skips deferred calls entirely.
 	if err := run(); err != nil {
@@ -24,9 +67,27 @@ func main() {
 	}
 }

+// storageDeps carries the engine-specific database handles and the repository
+// implementations. The usecases below only ever see the ports.
+type storageDeps struct {
+	tx             usecase.TxManager
+	taskRepo       usecase.TaskRepository
+	jobRepo        usecase.JobRepository
+	workerRepo     usecase.WorkerRepository
+	artifactRepo   usecase.ArtifactRepository
+	uiReadRepo     usecase.UIReadRepository
+	adminReadRepo  usecase.AdminReadRepository
+	settingsRepo   usecase.WorkloadSettingsRepository
+	taskResultRepo usecase.TaskResultRepository
+	statsRepo      interface {
+		Counts(ctx context.Context) (tasks, jobs, workers map[string]int, err error)
+	}
+	ready   func(ctx context.Context) error
+	migrate func(ctx context.Context, log *slog.Logger) error
+	close   func()
+}
+
 func run() error {
-	// Bootstrap logger, used only until config says where logs should go. It
-	// writes to stderr so it never contaminates the configured stdout stream.
 	boot := slog.New(slog.NewJSONHandler(os.Stderr, nil))

 	cfg, err := infra.LoadConfig()
@@ -34,6 +95,16 @@ func run() error {
 		boot.Error("load config", "err", err)
 		return err
 	}
+	return runWithConfig(cfg)
+}
+
+// runWithConfig boots the coordinator server with an explicit config. The
+// `serve` subcommand builds such a config for the single-binary mode; the
+// plain `coordinator` binary loads it from the environment.
+func runWithConfig(cfg infra.Config) error {
+	// Bootstrap logger, used only until config says where logs should go. It
+	// writes to stderr so it never contaminates the configured stdout stream.
+	boot := slog.New(slog.NewJSONHandler(os.Stderr, nil))

 	// The real logger: stdout plus an optional rotated file (LOG_FILE).
 	log, logCloser, err := infra.NewLogger(cfg)
@@ -46,12 +117,29 @@ func run() error {
 	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
 	defer stop()

-	pool, err := infra.NewPool(ctx, cfg, log)
+	var deps *storageDeps
+	switch cfg.DatabaseEngine {
+	case "sqlite":
+		deps, err = openSQLite(ctx, cfg, log)
+	case "postgres":
+		deps, err = openPostgres(ctx, cfg, log)
+	default:
+		err = fmt.Errorf("SCIMESH_DB must be sqlite or postgres")
+	}
 	if err != nil {
-		log.Error("connect database", "err", err)
+		log.Error("init storage", "err", err)
 		return err
 	}
-	defer pool.Close()
+	defer deps.close()
+
+	// A downloaded binary provisions its own schema; AUTO_MIGRATE=false keeps
+	// out-of-band migration workflows (the migrate CLI, CI, managed databases).
+	if cfg.AutoMigrate {
+		if err := deps.migrate(ctx, log); err != nil {
+			log.Error("apply migrations", "err", err)
+			return err
+		}
+	}

 	blobStore, err := blob.NewFSStore(cfg.StorageDir)
 	if err != nil {
@@ -59,40 +147,55 @@ func run() error {
 		return err
 	}

-	var (
-		clk            = infra.NewClock()
-		tx             = postgres.NewTxManager(pool)
-		taskRepo       = postgres.NewTaskRepo(pool)
-		jobRepo        = postgres.NewJobRepo(pool)
-		workerRepo     = postgres.NewWorkerRepo(pool)
-		artifactRepo   = postgres.NewArtifactRepo(pool)
-		uiReadRepo     = postgres.NewUIReadRepo(pool)
-		taskResultRepo = postgres.NewTaskResultRepo(pool)
-	)
+	clk := infra.NewClock()
+	tx, taskRepo, jobRepo, workerRepo, artifactRepo, uiReadRepo, taskResultRepo :=
+		deps.tx, deps.taskRepo, deps.jobRepo, deps.workerRepo, deps.artifactRepo, deps.uiReadRepo, deps.taskResultRepo
+
+	catalog, err := workloads.Load()
+	if err != nil {
+		log.Error("load workload catalog", "err", err)
+		return err
+	}

 	useCases := httptransport.UseCases{
 		RegisterWorker:   usecase.NewRegisterWorker(workerRepo, clk),
 		CreateJob:        usecase.NewCreateJob(jobRepo, taskRepo, tx, clk),
-		SubmitDataset:    usecase.NewSubmitDataset(blobStore, artifactRepo, jobRepo, taskRepo, tx, clk, cfg.DefaultMaxAttempts),
-		ClaimTask:        usecase.NewClaimTask(taskRepo, jobRepo, workerRepo, tx, clk, cfg.LeaseDuration),
+		SubmitDataset:    usecase.NewSubmitDataset(blobStore, artifactRepo, jobRepo, taskRepo, tx, clk, cfg.DefaultMaxAttempts, catalog, deps.settingsRepo),
+		ClaimTask:        usecase.NewClaimTask(taskRepo, jobRepo, workerRepo, tx, clk, cfg.LeaseDuration, catalog),
 		RenewLease:       usecase.NewRenewLease(taskRepo, workerRepo, tx, clk, cfg.LeaseDuration),
-		CompleteTask:     usecase.NewCompleteTask(taskRepo, jobRepo, artifactRepo, workerRepo, taskResultRepo, tx, clk, cfg.QuorumSize),
-		ReduceJob:        usecase.NewReduceJob(jobRepo, taskRepo, artifactRepo, blobStore, tx, clk),
-		FailTask:         usecase.NewFailTask(taskRepo, jobRepo, tx, clk),
+		CompleteTask:     usecase.NewCompleteTask(taskRepo, jobRepo, artifactRepo, workerRepo, taskResultRepo, tx, clk, cfg.QuorumSize, catalog),
+		ReduceJob:        usecase.NewReduceJob(jobRepo, taskRepo, artifactRepo, blobStore, tx, clk, catalog),
+		FailTask:         usecase.NewFailTask(taskRepo, jobRepo, workerRepo, tx, clk, catalog),
 		GetJobStatus:     usecase.NewGetJobStatus(jobRepo, taskRepo),
 		CancelJob:        usecase.NewCancelJob(jobRepo, taskRepo, tx, clk),
-		UploadArtifact:   usecase.NewUploadArtifact(taskRepo, artifactRepo, blobStore, tx, clk),
+		UploadArtifact:   usecase.NewUploadArtifact(taskRepo, workerRepo, artifactRepo, blobStore, tx, clk),
 		DownloadArtifact: usecase.NewDownloadArtifact(artifactRepo, blobStore),
 		GetJobResult:     usecase.NewGetJobResult(jobRepo, usecase.NewDownloadArtifact(artifactRepo, blobStore)),
 		GetTaskInput:     usecase.NewGetTaskInput(taskRepo, artifactRepo, blobStore),
-		Dashboard:        usecase.NewDashboard(uiReadRepo),
+		Dashboard:        usecase.NewDashboard(uiReadRepo, catalog),
+		PruneArtifacts:   usecase.NewPruneArtifacts(jobRepo, uiReadRepo, blobStore, clk),
 		PreviewArtifact:  usecase.NewPreviewArtifact(uiReadRepo, blobStore),
+		Admin: usecase.NewAdmin(deps.adminReadRepo, uiReadRepo, workerRepo, deps.settingsRepo, catalog,
+			usecase.AdminNodeInfo{
+				Version:     version,
+				StartedAt:   clk.Now(),
+				Binary:      executablePath(),
+				Addr:        cfg.Addr,
+				DataDir:     cfg.StorageDir,
+				DBEngine:    cfg.DatabaseEngine,
+				PublicURL:   cfg.PublicCoordinatorURL,
+				Userservice: cfg.UserserviceURL,
+				WorkerToken: func() string { return cfg.Token },
+			}, deps.ready, clk.Now).
+			WithAuditLog(log, func(ctx context.Context, action, detail string) {
+				log.Info("admin audit", "action", action, "detail", detail)
+			}),
 	}

 	// Background reapers are tracked so shutdown can wait for them. Without this
-	// the process would exit mid-UPDATE, and the deferred pool.Close() would pull
+	// the process would exit mid-UPDATE, and the deferred close() would pull
 	// connections out from under them.
-	expireLeases := usecase.NewExpireLeases(taskRepo, jobRepo, tx, clk)
+	expireLeases := usecase.NewExpireLeases(taskRepo, jobRepo, tx, clk, catalog)
 	markOffline := usecase.NewMarkWorkersOffline(workerRepo, clk, cfg.WorkerOfflineAfter)

 	var wg sync.WaitGroup
@@ -112,16 +215,15 @@ func run() error {

 	// Business metrics: gauges of tasks/jobs/workers by status, sampled from the
 	// database on every Prometheus scrape.
-	statsRepo := postgres.NewStatsRepo(pool)
 	m := metrics.New()
 	m.RegisterBusiness(func(ctx context.Context) (metrics.Stats, error) {
-		tasks, jobs, workers, err := statsRepo.Counts(ctx)
+		tasks, jobs, workers, err := deps.statsRepo.Counts(ctx)
 		return metrics.Stats{Tasks: tasks, Jobs: jobs, Workers: workers}, err
 	})

-	// pool.Ping backs /health: readiness means the database answers, not just
+	// deps.ready backs /health: readiness means the database answers, not just
 	// that the process is alive.
-	api := httptransport.NewServer(useCases, log, cfg.RequestTimeout, cfg.HeartbeatInterval, cfg.MaxUploadBytes, cfg.JWTSecret, cfg.UserserviceURL, m, pool.Ping, cfg.PublicCoordinatorURL, cfg.PublicUserserviceURL)
+	api := httptransport.NewServer(useCases, log, cfg.RequestTimeout, cfg.HeartbeatInterval, cfg.MaxUploadBytes, cfg.JWTSecret, cfg.UserserviceURL, m, deps.ready, cfg.PublicCoordinatorURL, cfg.PublicUserserviceURL, cfg.DocsDir)
 	err = infra.RunServer(ctx, log, cfg.Addr, api.Handler(cfg.Token, cfg.UIToken))

 	// Shutdown order matters, and defers alone cannot express it (they run
@@ -129,7 +231,7 @@ func run() error {
 	//
 	//   1. stop()    cancel the context, telling the reaper to finish
 	//   2. wg.Wait() let it return from its current tick
-	//   3. deferred pool.Close() closes an idle pool, not a busy one
+	//   3. deferred close() closes an idle pool, not a busy one
 	//
 	// Calling stop() here also covers the path where RunServer failed on its
 	// own: the context would never be cancelled otherwise and wg.Wait()
@@ -140,3 +242,64 @@ func run() error {

 	return err
 }
+
+// executablePath resolves the running binary for the admin console's node
+// information, falling back to the invocation name.
+func executablePath() string {
+	path, err := os.Executable()
+	if err != nil || path == "" {
+		return os.Args[0]
+	}
+	return path
+}
+
+// openSQLite opens the embedded database and builds the sqlite repositories.
+func openSQLite(ctx context.Context, cfg infra.Config, log *slog.Logger) (*storageDeps, error) {
+	if err := os.MkdirAll(cfg.StorageDir, 0o750); err != nil {
+		return nil, fmt.Errorf("create storage dir: %w", err)
+	}
+	db, err := sqlite.Open(cfg.DBPath)
+	if err != nil {
+		return nil, err
+	}
+	closeOnce := &sync.Once{}
+	return &storageDeps{
+		tx:             sqlite.NewTxManager(db),
+		taskRepo:       sqlite.NewTaskRepo(db),
+		jobRepo:        sqlite.NewJobRepo(db),
+		workerRepo:     sqlite.NewWorkerRepo(db),
+		artifactRepo:   sqlite.NewArtifactRepo(db),
+		uiReadRepo:     sqlite.NewUIReadRepo(db),
+		adminReadRepo:  sqlite.NewAdminReadRepo(db),
+		settingsRepo:   sqlite.NewWorkloadSettingsRepo(db),
+		taskResultRepo: sqlite.NewTaskResultRepo(db),
+		statsRepo:      sqlite.NewStatsRepo(db),
+		ready:          func(ctx context.Context) error { return db.PingContext(ctx) },
+		migrate:        func(ctx context.Context, log *slog.Logger) error { return sqlite.Migrate(ctx, db, log) },
+		close:          func() { closeOnce.Do(func() { _ = db.Close() }) },
+	}, nil
+}
+
+// openPostgres connects to PostgreSQL and builds the postgres repositories.
+func openPostgres(ctx context.Context, cfg infra.Config, log *slog.Logger) (*storageDeps, error) {
+	pool, err := infra.NewPool(ctx, cfg, log)
+	if err != nil {
+		return nil, err
+	}
+	closeOnce := &sync.Once{}
+	return &storageDeps{
+		tx:             postgres.NewTxManager(pool),
+		taskRepo:       postgres.NewTaskRepo(pool),
+		jobRepo:        postgres.NewJobRepo(pool),
+		workerRepo:     postgres.NewWorkerRepo(pool),
+		artifactRepo:   postgres.NewArtifactRepo(pool),
+		uiReadRepo:     postgres.NewUIReadRepo(pool),
+		adminReadRepo:  postgres.NewAdminReadRepo(pool),
+		settingsRepo:   postgres.NewWorkloadSettingsRepo(pool),
+		taskResultRepo: postgres.NewTaskResultRepo(pool),
+		statsRepo:      postgres.NewStatsRepo(pool),
+		ready:          func(ctx context.Context) error { return pool.Ping(ctx) },
+		migrate:        func(ctx context.Context, log *slog.Logger) error { return postgres.Migrate(ctx, cfg.DatabaseURL, log) },
+		close:          func() { closeOnce.Do(pool.Close) },
+	}, nil
+}
@@ -0,0 +1,339 @@
+package main
+
+import (
+	"context"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent"
+
+	"crypto/rand"
+	"encoding/hex"
+	"flag"
+	"fmt"
+	"log/slog"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
+	"strings"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/infra"
+	"github.com/emil28092005/SciMesh/coordinator/internal/userservice"
+)
+
+// runServe implements `coordinator serve`: the single-binary mode for a
+// scientist. It provisions a data directory (default ~/.scimesh) with the
+// coordinator and userservice sqlite databases, secrets, the admin account,
+// and optionally local worker agents — then runs the same server run() does.
+func runServe(args []string) error {
+	flags := flag.NewFlagSet("serve", flag.ContinueOnError)
+	flags.Usage = func() {
+		_, _ = fmt.Fprintf(flags.Output(), "usage: coordinator serve [options]\n")
+		_, _ = fmt.Fprintf(flags.Output(), "Runs the whole SciMesh platform from one binary: embedded databases, the\n")
+		_, _ = fmt.Fprintf(flags.Output(), "userservice, and optional local workers. No PostgreSQL or Docker needed.\n\n")
+		flags.PrintDefaults()
+	}
+	var (
+		dataDir   = flags.String("data-dir", defaultDataDir(), "data directory (default: ~/.scimesh)")
+		addr      = flags.String("addr", "127.0.0.1:8080", "listen address")
+		workers   = flags.Int("workers", 1, "number of local worker agents to spawn")
+		open      = flags.Bool("open", false, "open the UI in the browser")
+		docsDir   = flags.String("docs-dir", "", "built MkDocs site directory to serve at /ui/docs/")
+		email     = flags.String("admin-email", "admin@scimesh.local", "admin account email")
+		password  = flags.String("admin-password", "", "admin password (generated on first run when empty)")
+		publicURL = flags.String("public-url", "", "browser/worker-facing coordinator URL (default: http://<addr>)")
+	)
+	if err := flags.Parse(args); err != nil {
+		return err
+	}
+	if flags.NArg() > 0 {
+		return fmt.Errorf("serve takes no positional arguments")
+	}
+	if *workers < 0 {
+		return fmt.Errorf("--workers must be >= 0")
+	}
+
+	log := slog.New(slog.NewTextHandler(os.Stderr, nil))
+	if err := os.MkdirAll(*dataDir, 0o750); err != nil {
+		return fmt.Errorf("create data dir: %w", err)
+	}
+
+	// 1. Secrets, persisted in the data dir so restarts keep working.
+	jwtSecret, err := loadOrGenerate(filepath.Join(*dataDir, "jwt.secret"))
+	if err != nil {
+		return err
+	}
+	workerToken, err := loadOrGenerate(filepath.Join(*dataDir, "worker.token"))
+	if err != nil {
+		return err
+	}
+
+	// 2. Admin account: generated once and printed, remembered for later boots.
+	if *password == "" {
+		*password, err = loadOrGenerate(filepath.Join(*dataDir, "admin.password"))
+		if err != nil {
+			return err
+		}
+	}
+
+	// 3. Scientific runtime: ensure the managed venv (best effort).
+	venvPython := filepath.Join(*dataDir, "venv", binName("bin/python"))
+	ensureRuntime(log, *dataDir, venvPython)
+
+	// 4. Embedded userservice on the loopback interface.
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	usersAddr, closeUsers, err := userservice.Serve(ctx, userservice.Config{
+		DBPath:        filepath.Join(*dataDir, "users.db"),
+		JWTSecret:     jwtSecret,
+		AdminEmail:    *email,
+		AdminPassword: *password,
+		Log:           log,
+	})
+	if err != nil {
+		return fmt.Errorf("embedded userservice: %w", err)
+	}
+	defer func() { _ = closeUsers() }()
+
+	// 5. Local worker agents before the server, so they can claim immediately.
+	coordinatorURL := "http://" + *addr
+	agents, err := spawnAgents(ctx, log, *dataDir, *workers, coordinatorURL, workerToken, venvPython)
+	if err != nil {
+		return err
+	}
+	defer stopAgents(agents)
+
+	// 6. The coordinator server itself.
+	coordinatorPublicURL := *publicURL
+	if coordinatorPublicURL == "" {
+		coordinatorPublicURL = "http://" + *addr
+	}
+	cfg := infra.Config{
+		Addr:                 *addr,
+		DatabaseEngine:       "sqlite",
+		DBPath:               filepath.Join(*dataDir, "scimesh.db"),
+		Token:                workerToken,
+		JWTSecret:            jwtSecret,
+		UserserviceURL:       "http://" + usersAddr,
+		PublicCoordinatorURL: coordinatorPublicURL,
+		PublicUserserviceURL: "http://" + usersAddr,
+		LogLevel:             "info",
+		StorageDir:           filepath.Join(*dataDir, "artifacts"),
+		DocsDir:              *docsDir,
+		MaxUploadBytes:       1 << 30,
+		DBMaxConns:           4,
+		DBConnectTimeout:     10 * time.Second,
+		RequestTimeout:       15 * time.Second,
+		HeartbeatInterval:    15 * time.Second,
+		LeaseDuration:        2 * time.Minute,
+		DefaultMaxAttempts:   3,
+		QuorumSize:           2,
+		ReaperInterval:       30 * time.Second,
+		WorkerOfflineAfter:   1 * time.Minute,
+		AutoMigrate:          true,
+	}
+	if *open {
+		openBrowser("http://" + *addr + "/ui/admin")
+	}
+
+	// Print the login once the server is about to start.
+	fmt.Printf("\nSciMesh is starting at http://%s/ui\n", *addr)
+	fmt.Printf("  admin login: %s / %s\n", *email, *password)
+	if runtimeStatus(venvPython) {
+		fmt.Printf("  scientific runtime: ready (%s)\n", venvPython)
+	} else {
+		fmt.Printf("  scientific runtime: NOT ready — install Python 3, then restart serve\n")
+	}
+	fmt.Printf("  data directory: %s\n\n", *dataDir)
+
+	err = runWithConfig(cfg)
+	cancel()
+	stopAgents(agents)
+	return err
+}
+
+// defaultDataDir returns the platform-appropriate data directory.
+func defaultDataDir() string {
+	if dir := os.Getenv("SCIMESH_DATA_DIR"); dir != "" {
+		return dir
+	}
+	home, err := os.UserHomeDir()
+	if err != nil || home == "" {
+		return ".scimesh"
+	}
+	return filepath.Join(home, ".scimesh")
+}
+
+// loadOrGenerate reads a secret file, creating it with fresh random content
+// (chmod 0600) when missing.
+// #nosec G304 -- the path is an operator-supplied secret file inside the data dir.
+func loadOrGenerate(path string) (string, error) {
+	if raw, err := os.ReadFile(path); err == nil {
+		return strings.TrimSpace(string(raw)), nil
+	}
+	buffer := make([]byte, 32)
+	if _, err := rand.Read(buffer); err != nil {
+		return "", err
+	}
+	secret := hex.EncodeToString(buffer)
+	if err := os.WriteFile(path, []byte(secret+"\n"), 0o600); err != nil {
+		return "", fmt.Errorf("write %s: %w", path, err)
+	}
+	return secret, nil
+}
+
+// spawnAgents starts `coordinator agent` subprocesses that claim tasks from
+// the coordinator. Each gets its own work directory under the data dir.
+func spawnAgents(ctx context.Context, log *slog.Logger, dataDir string, count int,
+	coordinatorURL, token, venvPython string) ([]*exec.Cmd, error) {
+
+	var agents []*exec.Cmd
+	for i := 0; i < count; i++ {
+		workDir := filepath.Join(dataDir, "workers", fmt.Sprintf("%d", i))
+		if err := os.MkdirAll(workDir, 0o750); err != nil {
+			return agents, err
+		}
+		taskRunner := defaultTaskRunner(venvPython)
+		// #nosec G204,G702 -- the command is this binary itself with operator flags.
+		cmd := exec.CommandContext(ctx, os.Args[0], "agent",
+			"--coordinator-url", coordinatorURL,
+			"--token", token,
+			"--work-dir", workDir,
+			"--task-runner", taskRunner,
+		)
+		cmd.Stdout = os.Stdout
+		cmd.Stderr = os.Stderr
+		if err := cmd.Start(); err != nil {
+			return agents, fmt.Errorf("start local agent %d: %w", i, err)
+		}
+		agents = append(agents, cmd)
+		log.Info("local worker agent started", "index", i)
+	}
+	return agents, nil
+}
+
+// stopAgents terminates the spawned agents and waits briefly for them.
+func stopAgents(agents []*exec.Cmd) {
+	for _, agent := range agents {
+		if agent.Process != nil {
+			_ = agent.Process.Kill()
+		}
+	}
+	done := make(chan struct{})
+	go func() {
+		for _, agent := range agents {
+			_, _ = agent.Process.Wait()
+		}
+		close(done)
+	}()
+	select {
+	case <-done:
+	case <-time.After(5 * time.Second):
+	}
+}
+
+// defaultTaskRunner picks the managed venv python when present, else the
+// system `python`.
+func defaultTaskRunner(venvPython string) string {
+	if runtimeStatus(venvPython) {
+		return venvPython + " -m scimesh.worker.task"
+	}
+	return "python -m scimesh.worker.task"
+}
+
+// ensureRuntime creates the managed venv and installs scimesh into it, unless
+// it already exists. Best effort: a missing Python only logs a hint.
+func ensureRuntime(log *slog.Logger, dataDir, venvPython string) {
+	if runtimeStatus(venvPython) {
+		return
+	}
+	python := findPython()
+	if python == "" {
+		log.Warn("python3 not found; local workers need it to run scientific workloads")
+		return
+	}
+	log.Info("creating the scientific runtime venv", "python", python)
+	venvDir := filepath.Dir(filepath.Dir(venvPython))
+	// #nosec G204 -- python comes from PATH and venvDir from the data dir.
+	create := exec.CommandContext(context.Background(), python, "-m", "venv", venvDir)
+	if out, err := create.CombinedOutput(); err != nil {
+		log.Warn("venv creation failed; local workers need a manual Python install", "err", err, "output", string(out))
+		return
+	}
+	pip := filepath.Join(venvDir, binName("bin/pip"))
+	// The scimesh package is installed from an explicit source only: the PyPI
+	// name belongs to an unrelated project, so `pip install scimesh` would
+	// fetch a stranger's package. Default: download the wheel attached to our
+	// own GitHub release for this binary version; SCIMESH_PIP_PACKAGE
+	// overrides with a custom wheel, checkout or index.
+	source := os.Getenv("SCIMESH_PIP_PACKAGE")
+	if source == "" {
+		url, _, err := agent.ReleaseWheelURL(version)
+		if err != nil {
+			log.Warn("scientific runtime venv created, but scimesh is not installed",
+				"hint", "set SCIMESH_PIP_PACKAGE to your wheel or index")
+			return
+		}
+		downloaded, err := agent.DownloadWheel(context.Background(), url, venvDir)
+		if err != nil {
+			log.Warn("could not download the scimesh wheel for this release",
+				"err", err, "hint", "set SCIMESH_PIP_PACKAGE to your wheel or index")
+			return
+		}
+		source = downloaded
+	}
+	// #nosec G204,G702 -- pip and source are operator-configured paths.
+	install := exec.CommandContext(context.Background(), pip, "install", source)
+	if out, err := install.CombinedOutput(); err != nil {
+		log.Warn("pip install failed", "err", err, "output", string(out))
+		return
+	}
+	log.Info("scientific runtime installed", "venv", venvDir)
+}
+
+// findPython locates a usable python3.
+func findPython() string {
+	for _, candidate := range []string{"python3", "python"} {
+		path, err := exec.LookPath(candidate)
+		if err == nil {
+			return path
+		}
+	}
+	return ""
+}
+
+// runtimeStatus reports whether the managed venv python exists.
+func runtimeStatus(venvPython string) bool {
+	info, err := os.Stat(venvPython)
+	return err == nil && !info.IsDir()
+}
+
+// binName adapts a relative path to the platform layout.
+func binName(relative string) string {
+	if runtime.GOOS == "windows" {
+		parts := strings.Split(relative, "/")
+		parts[len(parts)-1] += ".exe"
+		return strings.Join(parts, string(filepath.Separator))
+	}
+	return relative
+}
+
+// openBrowser opens the UI in the platform's default browser.
+func openBrowser(target string) {
+	command := ""
+	switch runtime.GOOS {
+	case "darwin":
+		command = "open"
+	case "windows":
+		command = "rundll32"
+	default:
+		command = "xdg-open"
+	}
+	if command == "rundll32" {
+		// #nosec G204 -- target is the local UI URL the operator asked to open.
+		_ = exec.CommandContext(context.Background(), "rundll32", "url.dll,FileProtocolHandler", target).Start()
+		return
+	}
+	// #nosec G204 -- target is the local UI URL the operator asked to open.
+	_ = exec.CommandContext(context.Background(), command, target).Start()
+}
@@ -0,0 +1,74 @@
+package main
+
+import (
+	"context"
+	"flag"
+	"fmt"
+	"log/slog"
+	"os"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/setup"
+)
+
+// runSetup implements `coordinator setup` with non-interactive flags and an
+// interactive fallback for anything still missing.
+func runSetup(args []string) error {
+	flags := flag.NewFlagSet("setup", flag.ContinueOnError)
+	flags.Usage = func() {
+		_, _ = fmt.Fprintf(flags.Output(), "usage: coordinator setup [options]\n")
+		_, _ = fmt.Fprintf(flags.Output(), "Provisions the coordinator database, schema, and local .env settings.\n\n")
+		flags.PrintDefaults()
+	}
+	var (
+		databaseURL = flags.String("db", "", "coordinator database URL (default: DATABASE_URL)")
+		adminURL    = flags.String("admin-db", "", "maintenance URL to create a missing database (default: same host, 'postgres' db)")
+		envFile     = flags.String("env-file", "", "settings file to write (default: .env)")
+		force       = flags.Bool("force", false, "overwrite an existing settings file")
+		yes         = flags.Bool("yes", false, "non-interactive: use defaults, fail on anything missing")
+	)
+	if err := flags.Parse(args); err != nil {
+		return err
+	}
+	if flags.NArg() > 0 {
+		return fmt.Errorf("setup takes no positional arguments")
+	}
+
+	databaseURLValue := *databaseURL
+	if databaseURLValue == "" {
+		databaseURLValue = os.Getenv("DATABASE_URL")
+	}
+	envFileValue := *envFile
+	if envFileValue == "" {
+		envFileValue = os.Getenv("ENV_FILE")
+	}
+	adminURLValue := *adminURL
+	if adminURLValue == "" {
+		adminURLValue = os.Getenv("POSTGRES_ADMIN_URL")
+	}
+
+	options := setup.Options{
+		DatabaseURL:      databaseURLValue,
+		AdminDatabaseURL: adminURLValue,
+		EnvFile:          envFileValue,
+		Force:            *force,
+		Yes:              *yes,
+		ConnectTimeout:   10 * time.Second,
+		Out:              os.Stdout,
+		In:               os.Stdin,
+	}
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	log := slog.New(slog.NewTextHandler(os.Stderr, nil))
+	log.Info("setup started", "db", setup.SanitizeDatabaseURL(databaseURLValue), "env_file", envFileValue)
+
+	summary, err := setup.Run(ctx, options)
+	if err != nil {
+		log.Error("setup failed", "err", err)
+		return err
+	}
+	_, _ = fmt.Fprint(os.Stdout, summary)
+	log.Info("setup complete")
+	return nil
+}
@@ -0,0 +1,33 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"os"
+	"path/filepath"
+)
+
+// runToken implements `coordinator token`: prints the worker auth token of a
+// `coordinator serve` instance, so a scientist can join a worker without
+// hunting through the data directory. The file itself is what serve created.
+func runToken(args []string) error {
+	flags := flag.NewFlagSet("token", flag.ContinueOnError)
+	flags.Usage = func() {
+		_, _ = fmt.Fprintf(flags.Output(), "usage: coordinator token [options]\n")
+		_, _ = fmt.Fprintf(flags.Output(), "Prints the WORKER_AUTH_TOKEN of this coordinator's serve instance.\n\n")
+		flags.PrintDefaults()
+	}
+	var dataDir = flags.String("data-dir", defaultDataDir(), "data directory (default: ~/.scimesh)")
+	if err := flags.Parse(args); err != nil {
+		return err
+	}
+	if flags.NArg() > 0 {
+		return fmt.Errorf("token takes no positional arguments")
+	}
+	token, err := os.ReadFile(filepath.Join(*dataDir, "worker.token"))
+	if err != nil {
+		return fmt.Errorf("no worker token found in %s — start the coordinator with `coordinator serve` first", *dataDir)
+	}
+	fmt.Print(string(token))
+	return nil
+}
@@ -0,0 +1,238 @@
+// Command worker-agent is the Go worker agent: a coordinator client that
+// executes SDK workloads in a Python subprocess per claimed task. It also
+// carries the local setup wizard (`worker-agent setup`) so a machine that
+// installs only the worker can configure and start itself without a
+// coordinator on site.
+package main
+
+import (
+	"context"
+	"errors"
+	"flag"
+	"fmt"
+	"io"
+	"log/slog"
+	"net/http"
+	"os"
+	"os/exec"
+	"os/signal"
+	"strings"
+	"syscall"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent"
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent/setupui"
+)
+
+// version is injected at build time (-ldflags "-X main.version=...") and
+// reported by --version. "dev" marks a local build.
+var version = "dev"
+
+func main() {
+	// The wizard and --check need the injected build version too (they resolve
+	// the release wheel matching this binary), so it is set before dispatch.
+	agent.Version = version
+
+	if len(os.Args) > 1 && os.Args[1] == "setup" {
+		os.Exit(runSetup(os.Args[2:]))
+	}
+
+	fs := flag.NewFlagSet("worker-agent", flag.ExitOnError)
+	showVersion := fs.Bool("version", false, "print the build version and exit")
+	configPath := fs.String("config", "", "path to a JSON config file (SCIMESH_WORKER_CONFIG overrides the default)")
+	checkMode := fs.Bool("check", false, "run the preflight check (coordinator + local runtime) and exit 0/1")
+	checkURL := fs.String("coordinator-url", "", "coordinator URL to probe in --check mode")
+	_ = fs.Parse(os.Args[1:])
+
+	if *showVersion {
+		fmt.Println("worker-agent " + version)
+		return
+	}
+
+	if *checkMode {
+		url := *checkURL
+		if url == "" {
+			url = os.Getenv("COORDINATOR_URL")
+		}
+		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+		defer cancel()
+		if url == "" {
+			fmt.Println("check: no coordinator URL (pass --coordinator-url or set COORDINATOR_URL)")
+			os.Exit(1)
+		}
+		// Probe the managed venv when the wizard has installed it: workloads
+		// run with that interpreter, so checking the bare system python3
+		// would report a false negative.
+		checkPython, checkToken, checkKey, checkUsers := "", "", "", ""
+		if configPath := checkConfigPath(); configPath != "" {
+			checkPython = agent.VenvPython(configPath)
+			if config, err := agent.LoadConfigFile(configPath); err == nil {
+				checkToken, checkKey, checkUsers = config.Token, config.WorkerKey, config.UserserviceURL
+			}
+		}
+		report := agent.RunCheck(ctx, url, checkPython, checkToken, checkKey, checkUsers)
+		printCheck(report)
+		if !report.Coordinator.OK || !report.Python.OK || !report.Scimesh.OK {
+			os.Exit(1)
+		}
+		return
+	}
+
+	config, err := loadConfig(*configPath)
+	if err != nil {
+		slog.Error("invalid configuration", "error", err)
+		os.Exit(2)
+	}
+	logger := slog.New(slog.NewTextHandler(os.Stderr, nil))
+	tokens := agent.NewTokenProvider(
+		config.WorkerKey,
+		config.UserserviceURL,
+		config.Token,
+		config.RequestTimeout,
+	)
+	client := agent.NewClient(config.CoordinatorURL, tokens, config.RequestTimeout)
+	runner := agent.NewTaskRunner(config.TaskRunner)
+	daemon := agent.NewDaemon(config, client, runner, logger)
+	if err := daemon.RunForever(); err != nil {
+		logger.Error("agent stopped", "error", err)
+		os.Exit(1)
+	}
+}
+
+// checkConfigPath resolves where the wizard's config would be, honouring
+// SCIMESH_WORKER_CONFIG like the rest of the agent.
+func checkConfigPath() string {
+	if env := os.Getenv("SCIMESH_WORKER_CONFIG"); env != "" {
+		return env
+	}
+	return agent.DefaultConfigPath()
+}
+
+// loadConfig prefers a --config file; environment variables override the file
+// (see agent.ConfigFile.Config). Without a file, the plain environment path is
+// used exactly as before.
+func loadConfig(configPath string) (*agent.Config, error) {
+	if configPath != "" {
+		return agent.LoadConfigFile(configPath)
+	}
+	envPath := os.Getenv("SCIMESH_WORKER_CONFIG")
+	if envPath != "" {
+		if _, err := os.Stat(envPath); err == nil { //nolint:gosec // G703: path is the operator's own env var
+			return agent.LoadConfigFile(envPath)
+		}
+	}
+	return agent.LoadConfig()
+}
+
+func printCheck(report agent.CheckReport) {
+	fmt.Printf("worker-agent %s\n", report.Agent)
+	line := func(item agent.CheckItem) string {
+		mark := "✗"
+		if item.OK {
+			mark = "✓"
+		}
+		detail := item.Detail
+		if item.Latency > 0 {
+			detail = fmt.Sprintf("%s (%d ms)", detail, item.Latency)
+		}
+		return fmt.Sprintf("  %s %s: %s", mark, item.Name, detail)
+	}
+	fmt.Println(line(report.Coordinator))
+	fmt.Println(line(report.Auth))
+	fmt.Println(line(report.Python))
+	fmt.Println(line(report.Scimesh))
+}
+
+// runSetup serves the local setup wizard until interrupted. It binds the
+// loopback interface only.
+func runSetup(args []string) int {
+	fs := flag.NewFlagSet("worker-agent setup", flag.ContinueOnError)
+	port := fs.Int("port", 0, "listen port (default 12700)")
+	noOpen := fs.Bool("no-open", false, "do not open the browser automatically")
+	if err := fs.Parse(args); err != nil {
+		return 2
+	}
+	logger := slog.New(slog.NewTextHandler(os.Stderr, nil))
+
+	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
+	defer stop()
+
+	server := setupui.New(logger, setupui.Options{
+		Port: *port,
+		OpenBrowser: func(url string) {
+			if *noOpen {
+				return
+			}
+			openBrowser(url)
+		},
+	})
+	listener, err := server.Listen()
+	if err != nil {
+		// A wizard may already be running on this port (left open, or a
+		// second terminal). If it is ours, opening the browser is the
+		// friendlier outcome than failing the command.
+		if existing := wizardAlreadyRunning(*port); existing != "" {
+			logger.Info("the setup wizard is already running", "url", existing)
+			if !*noOpen {
+				openBrowser(existing)
+			}
+			return 0
+		}
+		logger.Error("setup wizard could not bind the loopback port", "err", err)
+		return 1
+	}
+	url := "http://" + listener.Addr().String()
+	logger.Info("SciMesh worker setup wizard", "url", url, "press-ctrl-c-to-stop", true)
+	server.OpenBrowser(url)
+	// Block until the signal arrives (never returns an error that matters: a
+	// cancelled context is the normal exit path).
+	err = server.Serve(ctx, listener)
+	if err != nil && !errors.Is(err, http.ErrServerClosed) {
+		logger.Error("setup wizard stopped", "err", err)
+		return 1
+	}
+	return 0
+}
+
+// openBrowser points the user's default browser at the wizard. Best-effort:
+// a missing browser must never fail the setup flow.
+func openBrowser(url string) {
+	for _, candidate := range [][]string{
+		{"xdg-open", url},
+		{"open", url},
+		{"cmd", "/c", "start", url},
+	} {
+		binary, err := exec.LookPath(candidate[0])
+		if err != nil {
+			continue
+		}
+		//nolint:gosec // G204: candidates are our own fixed list; the url is a loopback literal
+		_ = exec.CommandContext(context.Background(), binary, candidate[1:]...).Start()
+		return
+	}
+}
+
+// wizardAlreadyRunning probes the requested loopback port and returns its URL
+// when it serves the setup wizard page, or "" when it does not (another
+// process, or nothing at all).
+func wizardAlreadyRunning(port int) string {
+	url := fmt.Sprintf("http://127.0.0.1:%d/", port)
+	client := http.Client{Timeout: 2 * time.Second}
+	req, err := http.NewRequestWithContext(context.Background(), http.MethodGet, url, nil)
+	if err != nil {
+		return ""
+	}
+	resp, err := client.Do(req)
+	if err != nil {
+		return ""
+	}
+	defer func() { _ = resp.Body.Close() }()
+	if resp.StatusCode != http.StatusOK {
+		return ""
+	}
+	body, err := io.ReadAll(io.LimitReader(resp.Body, 1<<16))
+	if err != nil || !strings.Contains(string(body), "SciMesh Worker · Setup") {
+		return ""
+	}
+	return url
+}
@@ -20,20 +20,8 @@ services:
      retries: 10
      start_period: 5s

-  # One-shot: applies migrations, then exits. Schema changes stay an explicit
-  # deployment step — the coordinator binary never migrates on startup.
-  migrate:
-    image: migrate/migrate:v4.17.1
-    depends_on:
-      postgres:
-        condition: service_healthy
-    volumes:
-      - ./migrations:/migrations:ro
-    command:
-      - -path=/migrations
-      - -database=postgres://${POSTGRES_USER:-scimesh}:${POSTGRES_PASSWORD:-scimesh}@postgres:5432/${POSTGRES_DB:-scimesh}?sslmode=disable
-      - up
-    restart: on-failure
+  # The coordinator applies its embedded schema migrations on startup
+  # (AUTO_MIGRATE, on by default), so no separate migration step is needed.

  coordinator:
    build:
@@ -41,9 +29,6 @@ services:
    depends_on:
      postgres:
        condition: service_healthy
-      # Start only once the schema exists, otherwise the first query fails.
-      migrate:
-        condition: service_completed_successfully
    environment:
      COORDINATOR_ADDR: ":8080"
      # Host is the service name: compose resolves it on the project network.
@@ -51,6 +36,9 @@ services:
      WORKER_AUTH_TOKEN: ${WORKER_AUTH_TOKEN:-dev-token}
      # Empty disables /ui. Set this separately from the worker token.
      UI_AUTH_TOKEN: ${UI_AUTH_TOKEN:-}
+      # Directory of the built MkDocs site served at /ui/docs/ (empty disables
+      # the docs route; the demo mounts ./site automatically).
+      SCIMESH_DOCS_DIR: ${SCIMESH_DOCS_DIR:-}
      DB_MAX_CONNS: "10"
      REQUEST_TIMEOUT: "15s"
      LEASE_DURATION: "2m"
@@ -16,18 +16,26 @@ require (
 require (
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
+	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/jackc/pgpassfile v1.0.0 // indirect
 	github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a // indirect
 	github.com/jackc/puddle/v2 v2.2.1 // indirect
 	github.com/lann/builder v0.0.0-20180802200727-47ae307949d0 // indirect
 	github.com/lann/ps v0.0.0-20150810152359-62de8c46ede0 // indirect
+	github.com/mattn/go-isatty v0.0.20 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
+	github.com/ncruces/go-strftime v1.0.0 // indirect
 	github.com/prometheus/client_model v0.6.2 // indirect
 	github.com/prometheus/common v0.55.0 // indirect
 	github.com/prometheus/procfs v0.21.1 // indirect
+	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	golang.org/x/crypto v0.17.0 // indirect
 	golang.org/x/sync v0.22.0 // indirect
 	golang.org/x/sys v0.47.0 // indirect
 	golang.org/x/text v0.40.0 // indirect
 	google.golang.org/protobuf v1.36.11 // indirect
+	modernc.org/libc v1.74.1 // indirect
+	modernc.org/mathutil v1.7.1 // indirect
+	modernc.org/memory v1.11.0 // indirect
+	modernc.org/sqlite v1.55.0 // indirect
 )
@@ -9,6 +9,8 @@ github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XL
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
+github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
 github.com/golang-jwt/jwt/v5 v5.3.1 h1:kYf81DTWFe7t+1VvL7eS+jKFVWaUnK9cB1qbwn63YCY=
 github.com/golang-jwt/jwt/v5 v5.3.1/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArsqaEUEa5bE=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
@@ -29,8 +31,12 @@ github.com/lann/builder v0.0.0-20180802200727-47ae307949d0 h1:SOEGU9fKiNWd/HOJuq
 github.com/lann/builder v0.0.0-20180802200727-47ae307949d0/go.mod h1:dXGbAdH5GtBTC4WfIxhKZfyBF/HBFgRZSWwZ9g/He9o=
 github.com/lann/ps v0.0.0-20150810152359-62de8c46ede0 h1:P6pPBnrTSX3DEVR4fDembhRWSsG5rVo6hYhAB/ADZrk=
 github.com/lann/ps v0.0.0-20150810152359-62de8c46ede0/go.mod h1:vmVJ0l/dxyfGW6FmdpVm2joNMFikkuWg0EoCKLGUMNw=
+github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
+github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
+github.com/ncruces/go-strftime v1.0.0 h1:HMFp8mLCTPp341M/ZnA4qaf7ZlsbTc+miZjCLOFAw7w=
+github.com/ncruces/go-strftime v1.0.0/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/prometheus/client_golang v1.19.1 h1:wZWJDwK+NameRJuPGDhlnFgx8e8HN3XHQeLaYJFJBOE=
@@ -41,6 +47,8 @@ github.com/prometheus/common v0.55.0 h1:KEi6DK7lXW/m7Ig5i47x0vRzuBsHuvJdi5ee6Y3G
 github.com/prometheus/common v0.55.0/go.mod h1:2SECS4xJG1kd8XF9IcM1gMX6510RAEL65zxzNImwdc8=
 github.com/prometheus/procfs v0.21.1 h1:GljZCt+zSTS+NZq88cyQ1LjZ+RCHp3uVuabBWA5+OJI=
 github.com/prometheus/procfs v0.21.1/go.mod h1:aB55Cww9pdSJVHk0hUf0inxWyyjPogFIjmHKYgMKmtY=
+github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
+github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
 github.com/stretchr/testify v1.2.2/go.mod h1:a8OnRcib4nhh0OaRAV+Yts87kKdq0PP7pXfy6kDkUVs=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
@@ -51,6 +59,7 @@ golang.org/x/crypto v0.17.0 h1:r8bRNjWL3GshPW3gkd+RpvzWrZAwPS49OmTGZ/uhM4k=
 golang.org/x/crypto v0.17.0/go.mod h1:gCAAfMLgwOJRpTjQ2zCCt2OcSfYMTeZVSRtQlPC7Nq4=
 golang.org/x/sync v0.22.0 h1:SZjpbeLmrCk4xhRSZFNZW5gFUeCeFgjekvI/+gfScek=
 golang.org/x/sync v0.22.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
 golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/text v0.40.0 h1:Ub2Z6/xjgF1WrYQz2nuITOEegKFtiIy+rieRJ5lHZKs=
@@ -63,3 +72,11 @@ gopkg.in/natefinch/lumberjack.v2 v2.2.1/go.mod h1:YD8tP3GAjkrDg1eZH7EGmyESg/lsYs
 gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
+modernc.org/libc v1.74.1 h1:bdR4VTKFMC4966QSNZ05XLGI/VwzVa2kTUX51Dm0riQ=
+modernc.org/libc v1.74.1/go.mod h1:uH4t5bOx3G3g9Xcmj10YKlTcVISlRDwv8VoQJG9n8Os=
+modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
+modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
+modernc.org/memory v1.11.0 h1:o4QC8aMQzmcwCK3t3Ux/ZHmwFPzE6hf2Y5LbkRs+hbI=
+modernc.org/memory v1.11.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
+modernc.org/sqlite v1.55.0 h1:hIFh0MCH0rGinQ/4KYb5/UbCkRkb+UP+OkLCVWa5MTM=
+modernc.org/sqlite v1.55.0/go.mod h1:4ntCLuNmnH8+GNqjka1wNg7KJd5/Hi5FYp8K+XQ7GZw=
@@ -0,0 +1,121 @@
+package agent
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+)
+
+// TokenProvider supplies the current bearer token. A static token is served
+// forever; a worker key is exchanged at the userservice for short-lived JWTs
+// and refreshed before they expire (mirroring the former Python worker).
+type TokenProvider interface {
+	Token() (string, error)
+	Refresh() error
+}
+
+// StaticToken serves a fixed token forever; empty means no Authorization.
+type StaticToken struct{ token string }
+
+func (s *StaticToken) Token() (string, error) { return s.token, nil }
+func (s *StaticToken) Refresh() error         { return nil }
+
+// WorkerKeyToken exchanges a long-lived worker key for short-lived JWTs.
+type WorkerKeyToken struct {
+	userserviceURL string
+	workerKey      string
+	timeout        time.Duration
+	leeway         float64
+	mu             sync.Mutex
+	token          string
+	refreshAt      time.Time
+}
+
+func NewWorkerKeyToken(userserviceURL, workerKey string, timeout time.Duration) *WorkerKeyToken {
+	return &WorkerKeyToken{
+		userserviceURL: strings.TrimRight(userserviceURL, "/"),
+		workerKey:      workerKey,
+		timeout:        timeout,
+		leeway:         0.2,
+	}
+}
+
+// Token returns the current token, exchanging first when missing or stale.
+func (p *WorkerKeyToken) Token() (string, error) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	if p.token == "" || time.Now().After(p.refreshAt) {
+		if err := p.exchangeLocked(); err != nil {
+			return "", err
+		}
+	}
+	return p.token, nil
+}
+
+// Refresh forces an immediate exchange.
+func (p *WorkerKeyToken) Refresh() error {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	return p.exchangeLocked()
+}
+
+func (p *WorkerKeyToken) exchangeLocked() error {
+	payload, err := json.Marshal(map[string]string{"key": p.workerKey})
+	if err != nil {
+		return err
+	}
+	request, err := http.NewRequestWithContext(context.Background(), http.MethodPost, p.userserviceURL+"/worker-tokens/exchange", bytes.NewReader(payload))
+	if err != nil {
+		return err
+	}
+	request.Header.Set("Content-Type", "application/json")
+	client := &http.Client{Timeout: p.timeout}
+	response, err := client.Do(request)
+	if err != nil {
+		return fmt.Errorf("worker key exchange request failed")
+	}
+	defer func() { _ = response.Body.Close() }()
+	if response.StatusCode != http.StatusOK {
+		return fmt.Errorf("worker key exchange rejected with status %d", response.StatusCode)
+	}
+	raw, err := io.ReadAll(io.LimitReader(response.Body, 1<<20))
+	if err != nil {
+		return fmt.Errorf("worker key exchange request failed")
+	}
+	var data map[string]any
+	if err := json.Unmarshal(raw, &data); err != nil {
+		return fmt.Errorf("worker key exchange response is invalid")
+	}
+	token, _ := data["token"].(string)
+	if token == "" {
+		return fmt.Errorf("worker key exchange response is missing a token")
+	}
+	var ttl time.Duration
+	switch value := data["expires_in"].(type) {
+	case float64:
+		ttl = time.Duration(value * float64(time.Second))
+	case int:
+		ttl = time.Duration(value) * time.Second
+	}
+	p.token = token
+	p.refreshAt = time.Time{}
+	if ttl > 0 {
+		p.refreshAt = time.Now().Add(time.Duration(float64(ttl) * (1.0 - p.leeway)))
+	}
+	return nil
+}
+
+// NewTokenProvider picks the strategy: a worker key (with userservice) wins
+// over a static bearer token.
+func NewTokenProvider(workerKey, userserviceURL, bearerToken string, timeout time.Duration) TokenProvider {
+	if workerKey != "" && userserviceURL != "" {
+		return NewWorkerKeyToken(userserviceURL, workerKey, timeout)
+	}
+	return &StaticToken{token: bearerToken}
+}
@@ -0,0 +1,99 @@
+package agent
+
+import (
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+)
+
+func TestWorkerKeyTokenExchangesAndCaches(t *testing.T) {
+	var exchanges atomic.Int64
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/worker-tokens/exchange" {
+			http.NotFound(w, r)
+			return
+		}
+		var payload map[string]string
+		if err := json.NewDecoder(r.Body).Decode(&payload); err != nil || payload["key"] != "scimesh_wk_live_x" {
+			http.Error(w, "bad key", http.StatusUnauthorized)
+			return
+		}
+		exchanges.Add(1)
+		writeJSON(w, http.StatusOK, map[string]any{
+			"token":      "jwt-1",
+			"expires_in": 100,
+		})
+	}))
+	defer server.Close()
+
+	provider := NewWorkerKeyToken(server.URL, "scimesh_wk_live_x", 5*time.Second)
+	token, err := provider.Token()
+	if err != nil || token != "jwt-1" {
+		t.Fatalf("token = %q, err = %v", token, err)
+	}
+	// The second call within the TTL reuses the cache.
+	again, err := provider.Token()
+	if err != nil || again != "jwt-1" {
+		t.Fatalf("cached token = %q, err = %v", again, err)
+	}
+	if exchanges.Load() != 1 {
+		t.Errorf("exchanges = %d, want 1", exchanges.Load())
+	}
+	// An explicit refresh re-exchanges.
+	if err := provider.Refresh(); err != nil {
+		t.Fatalf("Refresh: %v", err)
+	}
+	if exchanges.Load() != 2 {
+		t.Errorf("exchanges after refresh = %d, want 2", exchanges.Load())
+	}
+}
+
+func TestWorkerKeyTokenRejectsBadKey(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusUnauthorized)
+	}))
+	defer server.Close()
+
+	provider := NewWorkerKeyToken(server.URL, "bad", 5*time.Second)
+	if _, err := provider.Token(); err == nil {
+		t.Error("expected exchange failure for a rejected key")
+	}
+}
+
+func TestNewTokenProviderSelectsStrategy(t *testing.T) {
+	if _, ok := NewTokenProvider("", "", "static", time.Second).(*StaticToken); !ok {
+		t.Error("expected a static token provider")
+	}
+	if _, ok := NewTokenProvider("key", "http://users", "", time.Second).(*WorkerKeyToken); !ok {
+		t.Error("expected a worker-key provider")
+	}
+}
+
+func TestClientRefreshesTokenOnceOn401(t *testing.T) {
+	var attempts atomic.Int64
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		attempts.Add(1)
+		if attempts.Load() == 1 {
+			w.WriteHeader(http.StatusUnauthorized)
+			return
+		}
+		writeJSON(w, http.StatusOK, map[string]any{})
+	}))
+	defer server.Close()
+
+	provider := &StaticToken{token: "t"}
+	client := NewClient(server.URL, provider, 5*time.Second)
+	status, _, err := client.requestJSON(http.MethodGet, "/ok", map[string]any{})
+	if err != nil {
+		t.Fatalf("request: %v", err)
+	}
+	if status != http.StatusOK {
+		t.Errorf("status = %d", status)
+	}
+	if attempts.Load() != 2 {
+		t.Errorf("attempts = %d, want 2 (401 then retry)", attempts.Load())
+	}
+}
@@ -0,0 +1,213 @@
+package agent
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"os/exec"
+	"runtime"
+	"strings"
+	"time"
+
+	guuid "github.com/google/uuid"
+)
+
+// CheckItem is one line of the preflight report the setup wizard shows.
+type CheckItem struct {
+	Name    string `json:"name"`
+	OK      bool   `json:"ok"`
+	Detail  string `json:"detail,omitempty"`
+	Latency int64  `json:"latency_ms,omitempty"`
+}
+
+// CheckReport is the full preflight result of `worker-agent --check` and of
+// the wizard's test step.
+type CheckReport struct {
+	Coordinator        CheckItem `json:"coordinator"`
+	Auth               CheckItem `json:"auth"`
+	Python             CheckItem `json:"python"`
+	Scimesh            CheckItem `json:"scimesh"`
+	Agent              string    `json:"agent_version"`
+	CoordinatorVersion string    `json:"coordinator_version,omitempty"`
+}
+
+// checkHTTP runs one GET and reports reachability + latency, with a fallback
+// detail message when the server answers without JSON.
+func checkHTTP(ctx context.Context, url string, timeout time.Duration) (CheckItem, string) {
+	started := time.Now()
+	reqCtx, cancel := context.WithTimeout(ctx, timeout)
+	defer cancel()
+	req, err := http.NewRequestWithContext(reqCtx, http.MethodGet, url, nil)
+	if err != nil {
+		return CheckItem{Name: "coordinator", OK: false, Detail: "invalid URL"}, ""
+	}
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		detail := err.Error()
+		if strings.Contains(detail, "connection refused") {
+			detail = "no coordinator answering at this address"
+		}
+		return CheckItem{Name: "coordinator", OK: false, Detail: detail}, ""
+	}
+	defer func() { _ = resp.Body.Close() }()
+	version := ""
+	if resp.StatusCode == http.StatusOK {
+		var body struct {
+			Status string `json:"status"`
+		}
+		if err := json.NewDecoder(resp.Body).Decode(&body); err == nil && body.Status == "ok" {
+			return CheckItem{Name: "coordinator", OK: true, Latency: time.Since(started).Milliseconds()}, version
+		}
+	}
+	return CheckItem{Name: "coordinator", OK: false, Detail: fmt.Sprintf("HTTP %d", resp.StatusCode)}, version
+}
+
+// CheckCoordinator probes the coordinator's /health endpoint.
+func CheckCoordinator(ctx context.Context, url string, timeout time.Duration) CheckReport {
+	report := CheckReport{Agent: Version}
+	item, _ := checkHTTP(ctx, strings.TrimRight(url, "/")+"/health", timeout)
+	report.Coordinator = item
+	report.Auth = CheckItem{Name: "auth", OK: true, Detail: "no token configured — will be checked at registration"}
+	return report
+}
+
+// CheckEnvironment verifies the local runtime against the python3 found on
+// PATH.
+func CheckEnvironment(ctx context.Context) CheckReport {
+	python, err := exec.LookPath("python3")
+	if err != nil {
+		return CheckReport{Agent: Version, Python: CheckItem{Name: "python", OK: false, Detail: "python3 not found on PATH"}}
+	}
+	return CheckEnvironmentWithPython(ctx, python)
+}
+
+// CheckEnvironmentWithPython verifies the local runtime against a specific
+// interpreter — the wizard's managed venv python when the runtime installer
+// has created one, so the preflight reflects what the worker will actually
+// execute with.
+func CheckEnvironmentWithPython(ctx context.Context, python string) CheckReport {
+	report := CheckReport{Agent: Version, Python: CheckItem{Name: "python", OK: true, Detail: python}}
+	//nolint:gosec // G204: python is a resolved interpreter path, the argument list is constant
+	cmd := exec.CommandContext(ctx, python, "-c", "import scimesh; print(scimesh.__version__ if hasattr(scimesh, '__version__') else 'installed')")
+	out, err := cmd.Output()
+	if err != nil {
+		// The worker executes workloads by spawning scimesh's task runner, so
+		// the package is a hard requirement, not an optimisation. The PyPI
+		// name belongs to a different project, so the wizard installs from
+		// SCIMESH_PIP_PACKAGE instead of suggesting a bare pip install.
+		report.Scimesh = CheckItem{Name: "scimesh", OK: false, Detail: "the worker runs workloads through scimesh — install it from your wheel or index (SCIMESH_PIP_PACKAGE)"}
+		return report
+	}
+	report.Scimesh = CheckItem{Name: "scimesh", OK: true, Detail: strings.TrimSpace(string(out))}
+	return report
+}
+
+// RunCheck combines the coordinator probe and the local environment probe; it
+// is the body behind `worker-agent --check` and the wizard's test step. A
+// non-empty python overrides the interpreter probed for the scimesh package
+// (the managed venv after a runtime install).
+// RunCheck combines the coordinator probe, a credential probe and the local
+// environment probe; it is the body behind `worker-agent --check` and the
+// wizard's test step. A non-empty python overrides the interpreter probed for
+// the scimesh package (the managed venv after a runtime install).
+func RunCheck(ctx context.Context, coordinatorURL, python, token, workerKey, userserviceURL string) CheckReport {
+	report := CheckCoordinator(ctx, coordinatorURL, 15*time.Second)
+	report.Auth = CheckAuth(ctx, coordinatorURL, token, workerKey, userserviceURL)
+	var env CheckReport
+	if python != "" {
+		env = CheckEnvironmentWithPython(ctx, python)
+	} else {
+		env = CheckEnvironment(ctx)
+	}
+	report.Python = env.Python
+	report.Scimesh = env.Scimesh
+	return report
+}
+
+// Version is the agent build version; main injects it via -ldflags and the
+// setup wizard mirrors it into the report. "dev" marks a local build.
+var Version = "dev"
+
+// Platform is the host platform string shown on the wizard.
+func Platform() string { return runtime.GOOS + "/" + runtime.GOARCH }
+
+// CheckAuth verifies the configured credential against the coordinator
+// without mutating anything: with a worker key it first exchanges it at the
+// userservice for a short-lived JWT, then it probes /tasks/claim with a
+// throwaway worker id and no capabilities. A 401 anywhere means the
+// credential was rejected; any other status proves it was accepted.
+func CheckAuth(ctx context.Context, url, token, workerKey, userserviceURL string) CheckItem {
+	item := CheckItem{Name: "auth"}
+	if token == "" && workerKey == "" {
+		item.OK = true
+		item.Detail = "no credential configured — will be checked at registration"
+		return item
+	}
+	client := &http.Client{Timeout: 30 * time.Second}
+	if workerKey != "" && userserviceURL != "" {
+		payload, _ := json.Marshal(map[string]string{"key": workerKey})
+		req, err := http.NewRequestWithContext(ctx, http.MethodPost, strings.TrimRight(userserviceURL, "/")+"/worker-tokens/exchange", strings.NewReader(string(payload)))
+		if err != nil {
+			item.OK = false
+			item.Detail = "invalid userservice URL"
+			return item
+		}
+		req.Header.Set("Content-Type", "application/json")
+		resp, err := client.Do(req)
+		if err != nil {
+			item.OK = false
+			item.Detail = "userservice unreachable: " + err.Error()
+			return item
+		}
+		defer func() { _ = resp.Body.Close() }()
+		if resp.StatusCode == http.StatusUnauthorized {
+			item.OK = false
+			item.Detail = "worker key rejected by the userservice"
+			return item
+		}
+		if resp.StatusCode != http.StatusOK {
+			item.OK = false
+			item.Detail = fmt.Sprintf("userservice exchange: HTTP %d", resp.StatusCode)
+			return item
+		}
+		var exchanged struct {
+			Token string `json:"token"`
+		}
+		if err := json.NewDecoder(resp.Body).Decode(&exchanged); err != nil || exchanged.Token == "" {
+			item.OK = false
+			item.Detail = "userservice exchange returned no token"
+			return item
+		}
+		token = exchanged.Token
+	}
+	if token == "" {
+		item.OK = false
+		item.Detail = "no usable credential after the key exchange"
+		return item
+	}
+	payload, _ := json.Marshal(map[string]any{"worker_id": guuid.NewString(), "capabilities": []string{}})
+	req, err := http.NewRequestWithContext(ctx, http.MethodPost, strings.TrimRight(url, "/")+"/tasks/claim", strings.NewReader(string(payload)))
+	if err != nil {
+		item.OK = false
+		item.Detail = "invalid coordinator URL"
+		return item
+	}
+	req.Header.Set("Content-Type", "application/json")
+	req.Header.Set("Authorization", "Bearer "+token)
+	resp, err := client.Do(req)
+	if err != nil {
+		item.OK = false
+		item.Detail = "coordinator unreachable: " + err.Error()
+		return item
+	}
+	defer func() { _ = resp.Body.Close() }()
+	if resp.StatusCode == http.StatusUnauthorized {
+		item.OK = false
+		item.Detail = "token rejected by the coordinator"
+		return item
+	}
+	item.OK = true
+	item.Detail = "credential accepted"
+	return item
+}
@@ -0,0 +1,68 @@
+package agent
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+)
+
+func TestCheckAuthAcceptsToken(t *testing.T) {
+	stub := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/tasks/claim" && r.Header.Get("Authorization") == "Bearer good-token" {
+			w.WriteHeader(http.StatusNoContent)
+			return
+		}
+		w.WriteHeader(http.StatusUnauthorized)
+	}))
+	defer stub.Close()
+
+	item := CheckAuth(context.Background(), stub.URL, "good-token", "", "")
+	if !item.OK {
+		t.Errorf("good token: %+v", item)
+	}
+	item = CheckAuth(context.Background(), stub.URL, "bad-token", "", "")
+	if item.OK {
+		t.Error("bad token must fail")
+	}
+}
+
+func TestCheckAuthNoCredentialIsNotAFailure(t *testing.T) {
+	item := CheckAuth(context.Background(), "http://coord:8080", "", "", "")
+	if !item.OK {
+		t.Errorf("no credential must not fail the preflight: %+v", item)
+	}
+}
+
+func TestCheckAuthWorkerKeyExchange(t *testing.T) {
+	var exchanged bool
+	users := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/worker-tokens/exchange" {
+			t.Errorf("unexpected userservice path %q", r.URL.Path)
+		}
+		exchanged = true
+		_, _ = w.Write([]byte(`{"token":"jwt-after-exchange"}`))
+	}))
+	defer users.Close()
+	coord := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.Header.Get("Authorization") == "Bearer jwt-after-exchange" {
+			w.WriteHeader(http.StatusNoContent)
+			return
+		}
+		w.WriteHeader(http.StatusUnauthorized)
+	}))
+	defer coord.Close()
+
+	item := CheckAuth(context.Background(), coord.URL, "", "smk_key", users.URL)
+	if !item.OK || !exchanged {
+		t.Errorf("key exchange flow: %+v exchanged=%v", item, exchanged)
+	}
+
+	rejected := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusUnauthorized)
+	}))
+	defer rejected.Close()
+	if item := CheckAuth(context.Background(), coord.URL, "", "smk_bad", rejected.URL); item.OK {
+		t.Error("rejected key must fail the preflight")
+	}
+}
@@ -0,0 +1,374 @@
+package agent
+
+import (
+	"bytes"
+	"context"
+	"crypto/sha256"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"time"
+)
+
+// CoordinatorError is a non-retriable coordinator response.
+type CoordinatorError struct{ msg string }
+
+func (e *CoordinatorError) Error() string { return e.msg }
+
+// TransientError is a timeout, connection error, or 5xx response.
+type TransientError struct{ msg string }
+
+func (e *TransientError) Error() string { return e.msg }
+
+// ConflictError means the worker no longer owns the task lease.
+type ConflictError struct{ msg string }
+
+func (e *ConflictError) Error() string { return e.msg }
+
+// Client speaks the v1 worker contract over HTTP with a token provider.
+//
+// API calls never follow redirects (a redirect is a contract violation); the
+// artifact download follows redirects but strips the Authorization header on
+// cross-origin hops, matching the Python worker's SameOriginAuthRedirectHandler.
+// A 401 response refreshes the token exactly once and retries, so a lapsed JWT
+// does not fail an in-flight task.
+type Client struct {
+	baseURL   string
+	tokens    TokenProvider
+	timeout   time.Duration
+	apiClient *http.Client
+	dlClient  *http.Client
+}
+
+func NewClient(baseURL string, tokens TokenProvider, timeout time.Duration) *Client {
+	return &Client{
+		baseURL: strings.TrimRight(baseURL, "/"),
+		tokens:  tokens,
+		timeout: timeout,
+		apiClient: &http.Client{
+			Timeout:       timeout,
+			CheckRedirect: func(*http.Request, []*http.Request) error { return http.ErrUseLastResponse },
+		},
+		dlClient: &http.Client{
+			Timeout: timeout,
+			CheckRedirect: func(req *http.Request, via []*http.Request) error {
+				if len(via) >= 10 {
+					return fmt.Errorf("too many redirects")
+				}
+				// Go strips Authorization on cross-host redirects by default;
+				// strip it explicitly on any origin change to be safe.
+				if len(via) > 0 && origin(req.URL) != origin(via[0].URL) {
+					req.Header.Del("Authorization")
+				}
+				return nil
+			},
+		},
+	}
+}
+
+func origin(u *url.URL) string {
+	return u.Scheme + "://" + u.Host
+}
+
+func (c *Client) authHeaders() (map[string]string, error) {
+	token, err := c.tokens.Token()
+	if err != nil {
+		return nil, &CoordinatorError{msg: "token refresh failed: " + err.Error()}
+	}
+	if token == "" {
+		return map[string]string{}, nil
+	}
+	return map[string]string{"Authorization": "Bearer " + token}, nil
+}
+
+func (c *Client) refreshAndRetry() bool {
+	return c.tokens.Refresh() == nil
+}
+
+// Register advertises the worker and returns its identity and heartbeat policy.
+func (c *Client) Register(name string, capabilities []string, cpuCount int, memoryMB int) (*RegisteredWorker, error) {
+	payload := map[string]any{
+		"name":         name,
+		"capabilities": capabilities,
+		"cpu_count":    cpuCount,
+	}
+	if memoryMB > 0 {
+		payload["memory_mb"] = memoryMB
+	}
+	status, body, err := c.requestJSON("POST", "/workers/register", payload)
+	if err != nil {
+		return nil, err
+	}
+	if status != http.StatusCreated {
+		return nil, &CoordinatorError{msg: fmt.Sprintf("worker registration rejected with status %d", status)}
+	}
+	return ParseRegistered(body)
+}
+
+// Claim leases one compatible task, or returns nil when the queue is empty.
+func (c *Client) Claim(workerID string, capabilities []string) (*Task, error) {
+	status, body, err := c.requestJSON("POST", "/tasks/claim", map[string]any{
+		"worker_id":       workerID,
+		"capabilities":    capabilities,
+		"max_concurrency": 1,
+	})
+	if err != nil {
+		return nil, err
+	}
+	if status == http.StatusNoContent {
+		return nil, nil
+	}
+	if status != http.StatusOK {
+		return nil, &CoordinatorError{msg: fmt.Sprintf("unexpected claim status %d", status)}
+	}
+	return ParseTask(body)
+}
+
+// Heartbeat renews the lease and returns the new deadline.
+func (c *Client) Heartbeat(task *Task, workerID string) (time.Time, error) {
+	status, body, err := c.requestJSON("POST", "/tasks/"+task.TaskID+"/heartbeat", map[string]any{
+		"worker_id": workerID,
+		"attempt":   task.Attempt,
+	})
+	if err != nil {
+		return time.Time{}, err
+	}
+	if status != http.StatusOK {
+		if status == http.StatusConflict {
+			return time.Time{}, &ConflictError{msg: "heartbeat rejected because the task lease was lost"}
+		}
+		return time.Time{}, &CoordinatorError{msg: fmt.Sprintf("heartbeat rejected with status %d", status)}
+	}
+	raw, ok := body["lease_expires_at"].(string)
+	if !ok {
+		return time.Time{}, &CoordinatorError{msg: "heartbeat response is missing lease_expires_at"}
+	}
+	lease, err := time.Parse(time.RFC3339, raw)
+	if err != nil {
+		return time.Time{}, &CoordinatorError{msg: "heartbeat returned an invalid lease_expires_at"}
+	}
+	task.LeaseExpiresAt = lease
+	task.leaseExpiresRaw = raw
+	return lease, nil
+}
+
+// Submit completes a task with the uploaded coordinator-owned artifact.
+func (c *Client) Submit(task *Task, workerID string, uploaded *Uploaded, metrics map[string]any) error {
+	status, _, err := c.requestJSON("POST", "/tasks/"+task.TaskID+"/result", map[string]any{
+		"worker_id": workerID,
+		"attempt":   task.Attempt,
+		"result":    map[string]any{"artifact_id": uploaded.ArtifactID},
+		"metrics":   metrics,
+	})
+	if err != nil {
+		return err
+	}
+	if status != http.StatusOK && status != http.StatusCreated && status != http.StatusAccepted {
+		if status == http.StatusConflict {
+			return &ConflictError{msg: "result rejected because the task lease was lost"}
+		}
+		return &CoordinatorError{msg: fmt.Sprintf("result rejected with status %d", status)}
+	}
+	return nil
+}
+
+// Fail reports a sanitized failure.
+func (c *Client) Fail(task *Task, workerID string, code, message string, retryable bool) error {
+	status, _, err := c.requestJSON("POST", "/tasks/"+task.TaskID+"/failure", map[string]any{
+		"worker_id":     workerID,
+		"attempt":       task.Attempt,
+		"error_code":    code,
+		"error_message": message,
+		"retryable":     retryable,
+	})
+	if err != nil {
+		return err
+	}
+	if status != http.StatusOK && status != http.StatusCreated && status != http.StatusAccepted {
+		if status == http.StatusConflict {
+			return &ConflictError{msg: "failure rejected because the task lease was lost"}
+		}
+		return &CoordinatorError{msg: fmt.Sprintf("failure report rejected with status %d", status)}
+	}
+	return nil
+}
+
+// Download streams the task input to destination and returns its SHA-256.
+func (c *Client) Download(uri, destination string) (string, error) {
+	resolved, err := url.Parse(uri)
+	if err != nil {
+		return "", fmt.Errorf("invalid input URI: %w", err)
+	}
+	if !resolved.IsAbs() {
+		base, parseErr := url.Parse(c.baseURL)
+		if parseErr != nil {
+			return "", fmt.Errorf("invalid coordinator URL")
+		}
+		resolved = base.ResolveReference(resolved)
+	}
+	if resolved.Scheme != "http" && resolved.Scheme != "https" {
+		return "", fmt.Errorf("input URI must be an HTTP(S) URL")
+	}
+	request, err := http.NewRequestWithContext(context.Background(), http.MethodGet, resolved.String(), nil)
+	if err != nil {
+		return "", err
+	}
+	headers, err := c.authHeaders()
+	if err != nil {
+		return "", err
+	}
+	for name, value := range headers {
+		request.Header.Set(name, value)
+	}
+	response, err := c.dlClient.Do(request)
+	if err != nil {
+		return "", &TransientError{msg: "input download failed"}
+	}
+	defer func() { _ = response.Body.Close() }()
+	if response.StatusCode == http.StatusUnauthorized && c.refreshAndRetry() {
+		return c.Download(uri, destination)
+	}
+	if response.StatusCode != http.StatusOK {
+		return "", &CoordinatorError{msg: fmt.Sprintf("input download rejected with status %d", response.StatusCode)}
+	}
+	if err := os.MkdirAll(filepath.Dir(destination), 0o750); err != nil {
+		return "", err
+	}
+	// #nosec G304 -- destination is the worker's own attempt directory file.
+	target, err := os.OpenFile(destination, os.O_CREATE|os.O_TRUNC|os.O_WRONLY, 0o600)
+	if err != nil {
+		return "", err
+	}
+	digest := sha256.New()
+	_, copyErr := io.Copy(io.MultiWriter(target, digest), response.Body)
+	closeErr := target.Close()
+	if copyErr != nil {
+		_ = os.Remove(destination)
+		return "", &TransientError{msg: "input download interrupted"}
+	}
+	if closeErr != nil {
+		return "", closeErr
+	}
+	return hex.EncodeToString(digest.Sum(nil)), nil
+}
+
+// Upload streams a partial artifact and verifies the returned metadata.
+func (c *Client) Upload(task *Task, workerID string, path, contentType string) (*Uploaded, error) {
+	// #nosec G304 -- the upload path is this worker's own artifact file.
+	file, err := os.Open(path)
+	if err != nil {
+		return nil, err
+	}
+	info, err := file.Stat()
+	if err != nil {
+		_ = file.Close()
+		return nil, err
+	}
+	digest := sha256.New()
+	if _, err := io.Copy(digest, file); err != nil {
+		_ = file.Close()
+		return nil, err
+	}
+	if _, err := file.Seek(0, io.SeekStart); err != nil {
+		_ = file.Close()
+		return nil, err
+	}
+	localSHA := hex.EncodeToString(digest.Sum(nil))
+	uploadURL := c.baseURL + "/tasks/" + url.PathEscape(task.TaskID) + "/artifacts/" + url.PathEscape(filepath.Base(path))
+	request, err := http.NewRequestWithContext(context.Background(), http.MethodPut, uploadURL, file)
+	if err != nil {
+		_ = file.Close()
+		return nil, err
+	}
+	request.ContentLength = info.Size()
+	request.Header.Set("Content-Type", contentType)
+	request.Header.Set("X-Worker-ID", workerID)
+	request.Header.Set("X-Task-Attempt", strconv.Itoa(task.Attempt))
+	headers, err := c.authHeaders()
+	if err != nil {
+		_ = file.Close()
+		return nil, err
+	}
+	for name, value := range headers {
+		request.Header.Set(name, value)
+	}
+	response, err := c.apiClient.Do(request)
+	_ = file.Close()
+	if err != nil {
+		return nil, &TransientError{msg: "artifact upload failed"}
+	}
+	defer func() { _ = response.Body.Close() }()
+	raw, err := io.ReadAll(io.LimitReader(response.Body, 1<<20))
+	if err != nil {
+		return nil, &TransientError{msg: "artifact upload interrupted"}
+	}
+	if response.StatusCode == http.StatusConflict {
+		return nil, &ConflictError{msg: "artifact upload rejected because the task lease was lost"}
+	}
+	if response.StatusCode == http.StatusUnauthorized && c.refreshAndRetry() {
+		return c.Upload(task, workerID, path, contentType)
+	}
+	if response.StatusCode != http.StatusOK {
+		return nil, &CoordinatorError{msg: fmt.Sprintf("artifact upload rejected with status %d", response.StatusCode)}
+	}
+	var payload map[string]any
+	if err := json.Unmarshal(raw, &payload); err != nil {
+		return nil, &CoordinatorError{msg: "artifact upload returned invalid metadata"}
+	}
+	uploaded, err := ParseUploaded(payload)
+	if err != nil {
+		return nil, &CoordinatorError{msg: "artifact upload returned invalid metadata"}
+	}
+	if uploaded.SHA256 != localSHA || uploaded.SizeBytes != info.Size() {
+		return nil, &CoordinatorError{msg: "artifact upload metadata does not match local artifact"}
+	}
+	return uploaded, nil
+}
+
+func (c *Client) requestJSON(method, path string, payload any) (int, map[string]any, error) {
+	body, err := json.Marshal(payload)
+	if err != nil {
+		return 0, nil, err
+	}
+	request, err := http.NewRequestWithContext(context.Background(), method, c.baseURL+path, bytes.NewReader(body))
+	if err != nil {
+		return 0, nil, err
+	}
+	request.Header.Set("Content-Type", "application/json")
+	headers, err := c.authHeaders()
+	if err != nil {
+		return 0, nil, err
+	}
+	for name, value := range headers {
+		request.Header.Set(name, value)
+	}
+	response, err := c.apiClient.Do(request)
+	if err != nil {
+		return 0, nil, &TransientError{msg: "coordinator request failed"}
+	}
+	defer func() { _ = response.Body.Close() }()
+	raw, err := io.ReadAll(io.LimitReader(response.Body, 1<<20))
+	if err != nil {
+		return 0, nil, &TransientError{msg: "coordinator request interrupted"}
+	}
+	if response.StatusCode == http.StatusUnauthorized && c.refreshAndRetry() {
+		return c.requestJSON(method, path, payload)
+	}
+	if response.StatusCode >= 500 {
+		return response.StatusCode, nil, &TransientError{msg: fmt.Sprintf("coordinator returned %d", response.StatusCode)}
+	}
+	var decoded map[string]any
+	if len(raw) > 0 {
+		if err := json.Unmarshal(raw, &decoded); err != nil {
+			return response.StatusCode, nil, &CoordinatorError{msg: "coordinator returned invalid JSON"}
+		}
+	}
+	return response.StatusCode, decoded, nil
+}
@@ -0,0 +1,225 @@
+package agent
+
+import (
+	"crypto/sha256"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+)
+
+func newTestClient(t *testing.T, server *httptest.Server) *Client {
+	t.Helper()
+	return NewClient(server.URL, &StaticToken{token: "test-token"}, 5*time.Second)
+}
+
+func TestClientRegisterClaimHeartbeat(t *testing.T) {
+	var registered, claimed, heartbeated bool
+	var server *httptest.Server //nolint:staticcheck // the handler closure references server
+	server = httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.Header.Get("Authorization") != "Bearer test-token" {
+			http.Error(w, "missing token", http.StatusUnauthorized)
+			return
+		}
+		switch {
+		case r.Method == http.MethodPost && r.URL.Path == "/workers/register":
+			registered = true
+			writeJSON(w, http.StatusCreated, map[string]any{
+				"worker_id":                  "22222222-2222-4222-8222-222222222222",
+				"heartbeat_interval_seconds": 15,
+			})
+		case r.Method == http.MethodPost && r.URL.Path == "/tasks/claim":
+			claimed = true
+			writeJSON(w, http.StatusOK, validTaskPayload())
+		case r.Method == http.MethodPost && r.URL.Path == "/tasks/11111111-1111-4111-8111-111111111111/heartbeat":
+			heartbeated = true
+			writeJSON(w, http.StatusOK, map[string]any{
+				"lease_expires_at": time.Now().Add(2 * time.Minute).UTC().Format(time.RFC3339),
+			})
+		default:
+			http.NotFound(w, r)
+		}
+	}))
+	defer server.Close()
+	client := newTestClient(t, server)
+
+	registeredWorker, err := client.Register("test-worker", []string{"similarity-search"}, 2, 1024)
+	if err != nil {
+		t.Fatalf("Register: %v", err)
+	}
+	if registeredWorker.WorkerID != "22222222-2222-4222-8222-222222222222" {
+		t.Errorf("worker id = %q", registeredWorker.WorkerID)
+	}
+
+	task, err := client.Claim("22222222-2222-4222-8222-222222222222", []string{"similarity-search"})
+	if err != nil {
+		t.Fatalf("Claim: %v", err)
+	}
+	if task == nil || task.Workload != "similarity-search" {
+		t.Fatalf("claim = %+v", task)
+	}
+
+	renewed, err := client.Heartbeat(task, "22222222-2222-4222-8222-222222222222")
+	if err != nil {
+		t.Fatalf("Heartbeat: %v", err)
+	}
+	if renewed.Before(time.Now()) {
+		t.Error("renewed lease is in the past")
+	}
+	if !registered || !claimed || !heartbeated {
+		t.Error("some endpoints were not hit")
+	}
+}
+
+func TestClientClaimEmptyAndConflict(t *testing.T) {
+	var server *httptest.Server //nolint:staticcheck // the handler closure references server
+	server = httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/tasks/claim":
+			w.WriteHeader(http.StatusNoContent)
+		case "/tasks/11111111-1111-4111-8111-111111111111/heartbeat":
+			w.WriteHeader(http.StatusConflict)
+		default:
+			http.NotFound(w, r)
+		}
+	}))
+	defer server.Close()
+	client := newTestClient(t, server)
+
+	task, err := client.Claim("worker", []string{"similarity-search"})
+	if err != nil {
+		t.Fatalf("Claim: %v", err)
+	}
+	if task != nil {
+		t.Error("expected no task for 204")
+	}
+
+	claimed, err := ParseTask(validTaskPayload())
+	if err != nil {
+		t.Fatalf("ParseTask: %v", err)
+	}
+	if _, err := client.Heartbeat(claimed, "worker"); err == nil {
+		t.Error("expected conflict error")
+	} else if !errors.As(err, &conflictError) {
+		t.Errorf("error type = %T", err)
+	}
+}
+
+func TestClientUploadSubmitFail(t *testing.T) {
+	var uploadedPath string
+	var server *httptest.Server //nolint:staticcheck // the handler closure references server
+	server = httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch {
+		case r.Method == http.MethodPut && strings.HasPrefix(r.URL.Path, "/tasks/11111111-1111-4111-8111-111111111111/artifacts/"):
+			if r.Header.Get("X-Worker-ID") != "worker" || r.Header.Get("X-Task-Attempt") != "1" {
+				t.Errorf("missing identity headers: %+v", r.Header)
+			}
+			uploadedPath = r.URL.Path
+			writeJSON(w, http.StatusOK, map[string]any{
+				"artifact_id": "33333333-3333-4333-8333-333333333333",
+				"uri":         server.URL + "/artifacts/333/download",
+				"sha256":      sha256Of(t, "partial body"),
+				"size_bytes":  int64(len("partial body")),
+			})
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/result"):
+			writeJSON(w, http.StatusAccepted, map[string]any{})
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/failure"):
+			writeJSON(w, http.StatusAccepted, map[string]any{})
+		default:
+			http.NotFound(w, r)
+		}
+	}))
+	defer server.Close()
+	client := newTestClient(t, server)
+	task, _ := ParseTask(validTaskPayload())
+
+	dir := t.TempDir()
+	partial := filepath.Join(dir, "result.csv")
+	if err := os.WriteFile(partial, []byte("partial body"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	uploaded, err := client.Upload(task, "worker", partial, "text/csv")
+	if err != nil {
+		t.Fatalf("Upload: %v", err)
+	}
+	if uploaded.SizeBytes != int64(len("partial body")) {
+		t.Errorf("size = %d", uploaded.SizeBytes)
+	}
+	if !strings.Contains(uploadedPath, "result.csv") {
+		t.Errorf("upload path = %q", uploadedPath)
+	}
+
+	if err := client.Submit(task, "worker", uploaded, map[string]any{"rows": 1}); err != nil {
+		t.Fatalf("Submit: %v", err)
+	}
+	if err := client.Fail(task, "worker", "ValueError", "bad input", false); err != nil {
+		t.Fatalf("Fail: %v", err)
+	}
+}
+
+func TestClientDownloadVerifiesChecksumAndStripsAuthOnRedirect(t *testing.T) {
+	var redirectedAuth string
+	bucket := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		redirectedAuth = r.Header.Get("Authorization")
+		_, _ = w.Write([]byte("input bytes"))
+	}))
+	defer bucket.Close()
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/tasks/11111111-1111-4111-8111-111111111111/input" {
+			http.Redirect(w, r, bucket.URL+"/presigned", http.StatusFound)
+			return
+		}
+		http.NotFound(w, r)
+	}))
+	defer server.Close()
+	client := newTestClient(t, server)
+
+	destination := filepath.Join(t.TempDir(), "input")
+	digest, err := client.Download(server.URL+"/tasks/11111111-1111-4111-8111-111111111111/input", destination)
+	if err != nil {
+		t.Fatalf("Download: %v", err)
+	}
+	if digest != sha256Of(t, "input bytes") {
+		t.Errorf("digest = %q", digest)
+	}
+	if redirectedAuth != "" {
+		t.Error("Authorization must be stripped on the redirected download")
+	}
+}
+
+func TestSanitizeErrorMessageRedactsPaths(t *testing.T) {
+	message := SanitizeErrorMessage(
+		"failed at /home/alice/work/attempts/1/input and /private/secret.txt",
+		"/home/alice/work",
+	)
+	for _, forbidden := range []string{"/home/alice", "/private/secret.txt"} {
+		if strings.Contains(message, forbidden) {
+			t.Errorf("message leaks %q: %q", forbidden, message)
+		}
+	}
+	if !strings.Contains(message, "<worker-dir>") {
+		t.Errorf("work dir not redacted: %q", message)
+	}
+	long := SanitizeErrorMessage(strings.Repeat("x", 500), "/tmp")
+	if len(long) != 300 {
+		t.Errorf("truncated length = %d", len(long))
+	}
+}
+
+func writeJSON(w http.ResponseWriter, status int, payload any) {
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(status)
+	_ = json.NewEncoder(w).Encode(payload)
+}
+
+func sha256Of(t *testing.T, value string) string {
+	t.Helper()
+	digest := sha256.Sum256([]byte(value))
+	return fmt.Sprintf("%x", digest)
+}
@@ -0,0 +1,177 @@
+package agent
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/workloads"
+)
+
+// Config is read only from the environment, mirroring the former Python
+// worker's configuration surface.
+type Config struct {
+	CoordinatorURL string
+	Token          string
+	WorkerKey      string
+	UserserviceURL string
+	WorkerName     string
+	WorkerID       string // set after registration; overridable for tests
+	WorkDir        string
+	CPUCount       int
+	MemoryMB       int // 0 = not advertised
+	PollInterval   time.Duration
+	RequestTimeout time.Duration
+	Heartbeat      time.Duration
+	CleanupAfter   time.Duration // 0 = keep attempt directories
+	Capabilities   []string
+	TaskRunner     []string // command + args; defaults to python -m scimesh.worker.task
+	MaxTasks       int      // 0 = unlimited
+	ExitWhenIdle   bool
+}
+
+func envList(name string) ([]string, error) {
+	raw := os.Getenv(name)
+	if raw == "" {
+		return nil, nil
+	}
+	var items []string
+	if err := json.Unmarshal([]byte(raw), &items); err != nil {
+		return nil, fmt.Errorf("%s must be a JSON array", name)
+	}
+	for _, item := range items {
+		if strings.TrimSpace(item) == "" {
+			return nil, fmt.Errorf("%s must not contain empty entries", name)
+		}
+	}
+	return items, nil
+}
+
+// LoadConfig validates the environment and fails fast on invalid values.
+func LoadConfig() (*Config, error) {
+	url := os.Getenv("COORDINATOR_URL")
+	if url == "" {
+		return nil, fmt.Errorf("COORDINATOR_URL is required")
+	}
+	if !strings.HasPrefix(url, "http://") && !strings.HasPrefix(url, "https://") {
+		return nil, fmt.Errorf("COORDINATOR_URL must be an absolute HTTP(S) URL")
+	}
+	workDir := os.Getenv("WORK_DIR")
+	if workDir == "" {
+		workDir = "./scimesh-agent-data"
+	}
+	cpu := 1
+	if raw := os.Getenv("CPU_COUNT"); raw != "" {
+		parsed, err := strconv.Atoi(raw)
+		if err != nil || parsed < 1 {
+			return nil, fmt.Errorf("CPU_COUNT must be a positive integer")
+		}
+		cpu = parsed
+	}
+	memoryMB := 0
+	if raw := os.Getenv("MEMORY_MB"); raw != "" {
+		parsed, err := strconv.Atoi(raw)
+		if err != nil || parsed < 1 {
+			return nil, fmt.Errorf("MEMORY_MB must be a positive integer")
+		}
+		memoryMB = parsed
+	}
+	poll, err := durationEnv("POLL_INTERVAL", 2*time.Second)
+	if err != nil {
+		return nil, err
+	}
+	timeout, err := durationEnv("REQUEST_TIMEOUT", 30*time.Second)
+	if err != nil {
+		return nil, err
+	}
+	heartbeat, err := durationEnv("HEARTBEAT_INTERVAL", 15*time.Second)
+	if err != nil {
+		return nil, err
+	}
+	cleanup, err := durationEnv("CLEANUP_AFTER_SECONDS", 0)
+	if err != nil {
+		return nil, err
+	}
+	capabilities, err := envList("CAPABILITIES")
+	if err != nil {
+		return nil, err
+	}
+	if len(capabilities) == 0 {
+		capabilities = DefaultCapabilities()
+	}
+	runner, err := envList("TASK_RUNNER")
+	if err != nil {
+		return nil, err
+	}
+	if len(runner) == 0 {
+		runner = []string{"python", "-m", "scimesh.worker.task"}
+	}
+	maxTasks := 0
+	if raw := os.Getenv("MAX_TASKS"); raw != "" {
+		parsed, err := strconv.Atoi(raw)
+		if err != nil || parsed < 1 {
+			return nil, fmt.Errorf("MAX_TASKS must be a positive integer")
+		}
+		maxTasks = parsed
+	}
+	name := os.Getenv("WORKER_NAME")
+	if name == "" {
+		host, _ := os.Hostname()
+		name = host
+	}
+	absWorkDir, err := filepath.Abs(workDir)
+	if err != nil {
+		return nil, fmt.Errorf("WORK_DIR must be an absolute path")
+	}
+	return &Config{
+		CoordinatorURL: strings.TrimRight(url, "/"),
+		Token:          os.Getenv("WORKER_AUTH_TOKEN"),
+		WorkerKey:      os.Getenv("WORKER_KEY"),
+		UserserviceURL: strings.TrimRight(os.Getenv("USERSERVICE_URL"), "/"),
+		WorkerName:     name,
+		WorkerID:       os.Getenv("WORKER_ID"),
+		WorkDir:        absWorkDir,
+		CPUCount:       cpu,
+		MemoryMB:       memoryMB,
+		PollInterval:   poll,
+		RequestTimeout: timeout,
+		Heartbeat:      heartbeat,
+		CleanupAfter:   cleanup,
+		Capabilities:   capabilities,
+		TaskRunner:     runner,
+		MaxTasks:       maxTasks,
+		ExitWhenIdle:   os.Getenv("EXIT_WHEN_IDLE") == "1",
+	}, nil
+}
+
+func durationEnv(name string, fallback time.Duration) (time.Duration, error) {
+	raw := os.Getenv(name)
+	if raw == "" {
+		return fallback, nil
+	}
+	parsed, err := time.ParseDuration(raw)
+	if err != nil || parsed < 0 {
+		return 0, fmt.Errorf("%s must be a non-negative duration", name)
+	}
+	return parsed, nil
+}
+
+// DefaultCapabilities derives the worker's advertised capabilities from the
+// embedded workload catalog, so an agent is workload-agnostic out of the box:
+// it claims whatever enabled workloads the coordinator library declares.
+// Explicit CAPABILITIES still overrides this for operators who want a subset.
+func DefaultCapabilities() []string {
+	catalog, err := workloads.Load()
+	if err != nil {
+		return []string{"similarity-search"}
+	}
+	names := make([]string, 0, len(catalog.Enabled()))
+	for _, workload := range catalog.Enabled() {
+		names = append(names, workload.Name)
+	}
+	return names
+}
@@ -0,0 +1,151 @@
+package agent
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+)
+
+// ConfigFile is the persisted worker configuration written by the setup
+// wizard and read back by `worker-agent --config`. Environment variables
+// still win: the file fills in what the environment left unset.
+type ConfigFile struct {
+	CoordinatorURL string   `json:"coordinator_url"`
+	Token          string   `json:"token,omitempty"`
+	WorkerKey      string   `json:"worker_key,omitempty"`
+	UserserviceURL string   `json:"userservice_url,omitempty"`
+	WorkDir        string   `json:"work_dir"`
+	WorkerName     string   `json:"worker_name,omitempty"`
+	CPUCount       int      `json:"cpu_count"`
+	MemoryMB       int      `json:"memory_mb"`
+	TaskRunner     []string `json:"task_runner,omitempty"`
+}
+
+// DefaultConfigPath is where the setup wizard stores the worker's
+// configuration. SCIMESH_WORKER_CONFIG overrides it.
+func DefaultConfigPath() string {
+	if raw := os.Getenv("SCIMESH_WORKER_CONFIG"); raw != "" {
+		return raw
+	}
+	home, err := os.UserHomeDir()
+	if err != nil || home == "" {
+		return filepath.Join(".", ".scimesh-worker", "config.json")
+	}
+	return filepath.Join(home, ".scimesh-worker", "config.json")
+}
+
+// LoadConfigFile reads and validates a persisted configuration. The file is
+// created by the wizard with 0600 permissions, so no credential is exposed to
+// other local users.
+func LoadConfigFile(path string) (*Config, error) {
+	//nolint:gosec // G304: path is --config or SCIMESH_WORKER_CONFIG, operator-supplied
+	raw, err := os.ReadFile(path)
+	if err != nil {
+		return nil, fmt.Errorf("read config file: %w", err)
+	}
+	var file ConfigFile
+	if err := json.Unmarshal(raw, &file); err != nil {
+		return nil, fmt.Errorf("parse config file: %w", err)
+	}
+	return file.Config()
+}
+
+// Config turns the file into the daemon configuration. Environment variables
+// take precedence so operators can still override any value per-process.
+func (f *ConfigFile) Config() (*Config, error) {
+	config := &Config{}
+	if env := os.Getenv("COORDINATOR_URL"); env != "" {
+		config.CoordinatorURL = env
+	} else {
+		config.CoordinatorURL = strings.TrimSpace(f.CoordinatorURL)
+	}
+	if config.CoordinatorURL == "" {
+		return nil, fmt.Errorf("coordinator_url is required")
+	}
+	if !strings.HasPrefix(config.CoordinatorURL, "http://") && !strings.HasPrefix(config.CoordinatorURL, "https://") {
+		return nil, fmt.Errorf("coordinator_url must be an absolute HTTP(S) URL")
+	}
+	if env := os.Getenv("WORKER_AUTH_TOKEN"); env != "" {
+		config.Token = env
+	} else {
+		config.Token = f.Token
+	}
+	if env := os.Getenv("WORKER_KEY"); env != "" {
+		config.WorkerKey = env
+	} else {
+		config.WorkerKey = f.WorkerKey
+	}
+	if env := os.Getenv("USERSERVICE_URL"); env != "" {
+		config.UserserviceURL = env
+	} else {
+		config.UserserviceURL = f.UserserviceURL
+	}
+	if env := os.Getenv("WORK_DIR"); env != "" {
+		config.WorkDir = env
+	} else if f.WorkDir != "" {
+		config.WorkDir = f.WorkDir
+	} else {
+		config.WorkDir = "./scimesh-agent-data"
+	}
+	if env := os.Getenv("WORKER_NAME"); env != "" {
+		config.WorkerName = env
+	} else {
+		config.WorkerName = f.WorkerName
+	}
+	config.CPUCount = f.CPUCount
+	if config.CPUCount < 1 {
+		config.CPUCount = 1
+	}
+	config.MemoryMB = f.MemoryMB
+	if config.MemoryMB < 0 {
+		config.MemoryMB = 0
+	}
+	if len(f.TaskRunner) > 0 {
+		config.TaskRunner = f.TaskRunner
+	}
+	if len(config.TaskRunner) == 0 {
+		config.TaskRunner = []string{"python", "-m", "scimesh.worker.task"}
+	}
+	config.PollInterval = 2 * time.Second
+	config.RequestTimeout = 30 * time.Second
+	config.Heartbeat = 15 * time.Second
+	config.Capabilities = DefaultCapabilities()
+	return config, nil
+}
+
+// Save writes the configuration file, creating the parent directory and
+// restricting permissions to the owner.
+func SaveConfigFile(path string, file ConfigFile) error {
+	payload, err := json.MarshalIndent(file, "", "  ")
+	if err != nil {
+		return err
+	}
+	payload = append(payload, '\n')
+	if err := os.MkdirAll(filepath.Dir(path), 0o700); err != nil {
+		return fmt.Errorf("create config directory: %w", err)
+	}
+	if err := os.WriteFile(path, payload, 0o600); err != nil {
+		return fmt.Errorf("write config file: %w", err)
+	}
+	return nil
+}
+
+// VenvPython returns the managed venv python next to the given config file,
+// or "" when the runtime installer has not created one yet. Mirrors the
+// wizard's probe so `worker-agent --check` and the preflight agree on what
+// interpreter will actually execute workloads.
+func VenvPython(configPath string) string {
+	dir := filepath.Dir(configPath)
+	for _, candidate := range []string{
+		filepath.Join(dir, "venv", "bin", "python"),
+		filepath.Join(dir, "venv", "Scripts", "python.exe"),
+	} {
+		if info, err := os.Stat(candidate); err == nil && !info.IsDir() {
+			return candidate
+		}
+	}
+	return ""
+}
@@ -0,0 +1,337 @@
+package agent
+
+import (
+	"errors"
+	"fmt"
+	"log/slog"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+	"time"
+)
+
+// Outcome reports whether a claim was made and whether it completed.
+type Outcome struct {
+	Claimed   bool
+	Completed bool
+}
+
+// Daemon is the agent state machine: register, claim, execute via the Python
+// task runner, upload, and submit — mirroring the Python worker's lifecycle.
+// conflictError is the errors.As target for lease conflicts.
+var conflictError *ConflictError
+
+type Daemon struct {
+	config     *Config
+	client     *Client
+	runner     *TaskRunner
+	log        *slog.Logger
+	workerID   string
+	registered bool
+	completed  int
+	mu         sync.Mutex
+}
+
+func NewDaemon(config *Config, client *Client, runner *TaskRunner, log *slog.Logger) *Daemon {
+	return &Daemon{config: config, client: client, runner: runner, log: log}
+}
+
+// RunForever loops until interrupted, idle-exit, or max tasks.
+func (d *Daemon) RunForever() error {
+	failures := 0
+	for {
+		if !d.registered {
+			if err := d.register(); err != nil {
+				return err
+			}
+		}
+		d.cleanupExpiredDirectories()
+		outcome, err := d.runOnce()
+		if err != nil {
+			failures++
+			d.log.Warn("agent cycle failed", "error", err)
+			backoff := d.config.PollInterval
+			for i := 0; i < failures && i < 6; i++ {
+				backoff *= 2
+			}
+			if backoff > 60*time.Second {
+				backoff = 60 * time.Second
+			}
+			time.Sleep(backoff)
+			continue
+		}
+		failures = 0
+		if outcome.Claimed && outcome.Completed {
+			d.completed++
+			if d.config.MaxTasks > 0 && d.completed >= d.config.MaxTasks {
+				d.log.Info("max tasks reached")
+				return nil
+			}
+		}
+		if !outcome.Claimed && d.config.ExitWhenIdle {
+			d.log.Info("queue empty, exiting")
+			return nil
+		}
+		if outcome.Claimed && d.config.ExitWhenIdle {
+			d.log.Info("one claim processed, exiting")
+			return nil
+		}
+		if !outcome.Claimed {
+			time.Sleep(d.config.PollInterval)
+		}
+	}
+}
+
+func (d *Daemon) register() error {
+	registered, err := d.client.Register(
+		d.config.WorkerName,
+		d.config.Capabilities,
+		d.config.CPUCount,
+		d.config.MemoryMB,
+	)
+	if err != nil {
+		return err
+	}
+	d.mu.Lock()
+	if d.config.WorkerID != "" {
+		d.workerID = d.config.WorkerID
+	} else {
+		d.workerID = registered.WorkerID
+	}
+	d.registered = true
+	d.mu.Unlock()
+	d.log.Info("registered", "worker_id", d.workerID)
+	return nil
+}
+
+func (d *Daemon) workerIDOrEmpty() string {
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	return d.workerID
+}
+
+// cleanupExpiredDirectories removes task attempt directories older than the
+// configured retention, mirroring the former Python worker's cleanup.
+func (d *Daemon) cleanupExpiredDirectories() {
+	if d.config.CleanupAfter <= 0 {
+		return
+	}
+	cutoff := time.Now().Add(-d.config.CleanupAfter)
+	tasks, err := os.ReadDir(d.config.WorkDir)
+	if err != nil {
+		return
+	}
+	for _, taskEntry := range tasks {
+		if !taskEntry.IsDir() {
+			continue
+		}
+		taskDir := filepath.Join(d.config.WorkDir, taskEntry.Name())
+		attempts, err := os.ReadDir(taskDir)
+		if err != nil {
+			continue
+		}
+		for _, attemptEntry := range attempts {
+			if !attemptEntry.IsDir() {
+				continue
+			}
+			info, err := attemptEntry.Info()
+			if err == nil && info.ModTime().Before(cutoff) {
+				_ = os.RemoveAll(filepath.Join(taskDir, attemptEntry.Name()))
+			}
+		}
+		if entries, err := os.ReadDir(taskDir); err == nil && len(entries) == 0 {
+			_ = os.Remove(taskDir)
+		}
+	}
+}
+
+func (d *Daemon) runOnce() (Outcome, error) {
+	workerID := d.workerIDOrEmpty()
+	if workerID == "" {
+		return Outcome{}, fmt.Errorf("agent is not registered")
+	}
+	task, err := d.client.Claim(workerID, d.config.Capabilities)
+	if err != nil {
+		return Outcome{}, err
+	}
+	if task == nil {
+		return Outcome{Claimed: false}, nil
+	}
+	d.log.Info("task claimed", "task_id", task.TaskID, "attempt", task.Attempt)
+	started := time.Now()
+	taskDir := filepath.Join(d.config.WorkDir, task.TaskID, fmt.Sprint(task.Attempt))
+	if err := os.MkdirAll(taskDir, 0o750); err != nil {
+		return Outcome{Claimed: true}, err
+	}
+	heartbeat := newLeaseHeartbeat(task, workerID, d.client, d.config.Heartbeat)
+	completed := false
+	err = heartbeat.Start()
+	if err != nil {
+		if errors.As(err, &conflictError) {
+			d.log.Warn("lease lost", "task_id", task.TaskID)
+			return Outcome{Claimed: true}, nil
+		}
+		return Outcome{Claimed: true}, err
+	}
+	defer heartbeat.Stop()
+
+	// Attempt directory cleanup is deliberately minimal in the prototype:
+	// attempt directories are retained under the work directory.
+	failure := d.executeTask(task, workerID, taskDir, started, heartbeat)
+	if failure != nil {
+		if errors.As(failure, &conflictError) {
+			d.log.Warn("lease lost", "task_id", task.TaskID)
+			return Outcome{Claimed: true}, nil
+		}
+		if err := heartbeat.RaiseIfFailed(); err != nil {
+			return Outcome{Claimed: true}, nil
+		}
+		d.reportFailure(task, workerID, failure)
+		return Outcome{Claimed: true}, nil
+	}
+	if err := heartbeat.RaiseIfFailed(); err != nil {
+		return Outcome{Claimed: true}, nil
+	}
+	completed = true
+	d.log.Info("task completed", "task_id", task.TaskID, "elapsed_seconds", time.Since(started).Seconds())
+	return Outcome{Claimed: true, Completed: completed}, nil
+}
+
+// executeTask returns nil on success or a classified failure.
+func (d *Daemon) executeTask(task *Task, workerID, taskDir string, started time.Time, heartbeat *leaseHeartbeat) error {
+	inputPath := filepath.Join(taskDir, "input")
+	// Downloads use the coordinator-provided URI verbatim; a relative path is
+	// resolved against the coordinator by the client.
+	actualSHA, err := d.client.Download(task.Input.URI, inputPath)
+	if err != nil {
+		return err
+	}
+	if !strings.EqualFold(actualSHA, task.Input.SHA256) {
+		return &CoordinatorError{msg: "input checksum mismatch"}
+	}
+	if err := heartbeat.RaiseIfFailed(); err != nil {
+		return err
+	}
+	manifestPath := filepath.Join(taskDir, "manifest.json")
+	manifest, err := d.runner.Run(task, taskDir, manifestPath, nil)
+	if err != nil {
+		return err
+	}
+	if err := heartbeat.RaiseIfFailed(); err != nil {
+		return err
+	}
+	uploaded, err := d.client.Upload(task, workerID, manifest.ArtifactPath, manifest.ContentType)
+	if err != nil {
+		return err
+	}
+	if err := heartbeat.RaiseIfFailed(); err != nil {
+		return err
+	}
+	metrics := map[string]any{"elapsed_seconds": roundSeconds(time.Since(started).Seconds())}
+	for name, value := range manifest.Metrics {
+		metrics[name] = value
+	}
+	return d.client.Submit(task, workerID, uploaded, metrics)
+}
+
+func (d *Daemon) reportFailure(task *Task, workerID string, failure error) {
+	var code string
+	var coordinatorErr *CoordinatorError
+	if errors.As(failure, &coordinatorErr) {
+		code = "ValueError"
+	} else {
+		code = "TaskRunnerFailed"
+	}
+	retryable := IsRetryableError(failure)
+	message := SanitizeErrorMessage(failure.Error(), d.config.WorkDir)
+	d.log.Warn("task failed", "task_id", task.TaskID, "error_code", code, "retryable", retryable)
+	if err := d.client.Fail(task, workerID, code, message, retryable); err != nil {
+		if errors.As(err, &conflictError) {
+			d.log.Warn("lease lost while reporting failure", "task_id", task.TaskID)
+			return
+		}
+		d.log.Warn("failure report rejected", "task_id", task.TaskID, "error", err)
+	}
+}
+
+// leaseHeartbeat renews the lease from the returned deadline at less than
+// half of the remaining TTL, mirroring the Python worker.
+type leaseHeartbeat struct {
+	task     *Task
+	workerID string
+	client   *Client
+	interval time.Duration
+	stop     chan struct{}
+	once     sync.Once
+	mu       sync.Mutex
+	lease    time.Time
+	failed   error
+}
+
+func newLeaseHeartbeat(task *Task, workerID string, client *Client, interval time.Duration) *leaseHeartbeat {
+	return &leaseHeartbeat{
+		task:     task,
+		workerID: workerID,
+		client:   client,
+		interval: interval,
+		stop:     make(chan struct{}),
+		lease:    task.LeaseExpiresAt,
+	}
+}
+
+func (h *leaseHeartbeat) Start() error {
+	if _, err := h.client.Heartbeat(h.task, h.workerID); err != nil {
+		return err
+	}
+	go h.loop()
+	return nil
+}
+
+func (h *leaseHeartbeat) Stop() {
+	h.once.Do(func() { close(h.stop) })
+}
+
+func (h *leaseHeartbeat) RaiseIfFailed() error {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	return h.failed
+}
+
+func (h *leaseHeartbeat) loop() {
+	for {
+		delay := h.nextDelay()
+		select {
+		case <-h.stop:
+			return
+		case <-time.After(delay):
+		}
+		renewed, err := h.client.Heartbeat(h.task, h.workerID)
+		h.mu.Lock()
+		if err != nil {
+			h.failed = err
+			h.mu.Unlock()
+			return
+		}
+		h.lease = renewed
+		h.mu.Unlock()
+	}
+}
+
+func (h *leaseHeartbeat) nextDelay() time.Duration {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	remaining := time.Until(h.lease)
+	if remaining <= 0 {
+		return 0
+	}
+	half := remaining / 2
+	if h.interval < half {
+		return h.interval
+	}
+	return half
+}
+
+func roundSeconds(seconds float64) float64 {
+	return float64(int64(seconds*1000)) / 1000
+}
@@ -0,0 +1,296 @@
+package agent
+
+import (
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"log/slog"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+)
+
+func sha256HexOf(value string) string {
+	sum := sha256.Sum256([]byte(value))
+	return fmt.Sprintf("%x", sum)
+}
+
+// fakeRunnerScript writes a result manifest for --output and exits with the
+// given code.
+func fakeRunnerScript(t *testing.T, dir string, exitCode int) string {
+	t.Helper()
+	script := filepath.Join(dir, "fake-runner.sh")
+	content := `#!/bin/sh
+out=""
+task_dir=""
+while [ "$#" -gt 0 ]; do
+  case "$1" in
+    --output) out="$2"; shift 2;;
+    --task-dir) task_dir="$2"; shift 2;;
+    *) shift;;
+  esac
+done
+printf 'id,score\n' > "$task_dir/result.csv"
+printf '{"artifact_path":"%s/result.csv","content_type":"text/csv","metrics":{"rows":1}}' "$task_dir" > "$out"
+exit ` + fmt.Sprint(exitCode) + "\n"
+	if err := os.WriteFile(script, []byte(content), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	return script
+}
+
+// fakeCoordinator implements the v1 contract over HTTP and records calls.
+type fakeCoordinator struct {
+	server     *httptest.Server
+	task       map[string]any
+	submits    []map[string]any
+	failures   []map[string]any
+	heartbeats int
+	uploadSHA  string
+	uploadSize int64
+	inputBytes []byte
+	conflict   bool // 409 on heartbeat/upload/result
+}
+
+func newFakeCoordinator(t *testing.T, task map[string]any) *fakeCoordinator {
+	t.Helper()
+	fake := &fakeCoordinator{task: task, inputBytes: []byte("input fixture")}
+	fake.uploadSHA = sha256HexOf(string(fake.inputBytes))
+	fake.uploadSize = int64(len(fake.inputBytes))
+	var server *httptest.Server
+	server = httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch {
+		case r.Method == http.MethodPost && r.URL.Path == "/workers/register":
+			writeJSON(w, http.StatusCreated, map[string]any{
+				"worker_id":                  "22222222-2222-4222-8222-222222222222",
+				"heartbeat_interval_seconds": 15.0,
+			})
+		case r.Method == http.MethodPost && r.URL.Path == "/tasks/claim":
+			if fake.task == nil {
+				w.WriteHeader(http.StatusNoContent)
+				return
+			}
+			writeJSON(w, http.StatusOK, fake.task)
+		case r.Method == http.MethodGet && strings.HasSuffix(r.URL.Path, "/input"):
+			_, _ = w.Write(fake.inputBytes)
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/heartbeat"):
+			fake.heartbeats++
+			if fake.conflict {
+				w.WriteHeader(http.StatusConflict)
+				return
+			}
+			writeJSON(w, http.StatusOK, map[string]any{
+				"lease_expires_at": time.Now().Add(2 * time.Minute).UTC().Format(time.RFC3339),
+			})
+		case r.Method == http.MethodPut && strings.Contains(r.URL.Path, "/artifacts/"):
+			raw, _ := io.ReadAll(r.Body)
+			fake.uploadSize = int64(len(raw))
+			fake.uploadSHA = fmt.Sprintf("%x", sha256.Sum256(raw))
+			if fake.conflict {
+				w.WriteHeader(http.StatusConflict)
+				return
+			}
+			writeJSON(w, http.StatusOK, map[string]any{
+				"artifact_id": "33333333-3333-4333-8333-333333333333",
+				"uri":         server.URL + "/artifacts/333/download",
+				"sha256":      fake.uploadSHA,
+				"size_bytes":  fake.uploadSize,
+			})
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/result"):
+			var payload map[string]any
+			_ = json.NewDecoder(r.Body).Decode(&payload)
+			fake.submits = append(fake.submits, payload)
+			if fake.conflict {
+				w.WriteHeader(http.StatusConflict)
+				return
+			}
+			w.WriteHeader(http.StatusAccepted)
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/failure"):
+			var payload map[string]any
+			_ = json.NewDecoder(r.Body).Decode(&payload)
+			fake.failures = append(fake.failures, payload)
+			w.WriteHeader(http.StatusAccepted)
+		default:
+			http.NotFound(w, r)
+		}
+	}))
+	fake.server = server
+	return fake
+}
+
+func (f *fakeCoordinator) close() { f.server.Close() }
+
+func testDaemon(t *testing.T, fake *fakeCoordinator, script string) *Daemon {
+	t.Helper()
+	config := &Config{
+		CoordinatorURL: fake.server.URL,
+		WorkerName:     "test-worker",
+		WorkerID:       "22222222-2222-4222-8222-222222222222",
+		WorkDir:        t.TempDir(),
+		CPUCount:       1,
+		PollInterval:   time.Millisecond,
+		RequestTimeout: 5 * time.Second,
+		Heartbeat:      15 * time.Second,
+		Capabilities:   []string{"similarity-search"},
+		TaskRunner:     []string{script},
+	}
+	client := NewClient(fake.server.URL, &StaticToken{token: "test-token"}, 5*time.Second)
+	runner := NewTaskRunner(config.TaskRunner)
+	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
+	daemon := NewDaemon(config, client, runner, logger)
+	if err := daemon.register(); err != nil {
+		t.Fatalf("register: %v", err)
+	}
+	return daemon
+}
+
+func validClaimedTaskPayload() map[string]any {
+	return map[string]any{
+		"task_id":          "11111111-1111-4111-8111-111111111111",
+		"attempt":          1.0,
+		"lease_expires_at": time.Now().Add(time.Minute).UTC().Format(time.RFC3339),
+		"workload":         "similarity-search",
+		"input": map[string]any{
+			"uri":    "/tasks/11111111-1111-4111-8111-111111111111/input",
+			"sha256": sha256HexOf("input fixture"),
+		},
+		"parameters": map[string]any{"query_smiles": "CCO"},
+	}
+}
+
+func TestDaemonCompletesAClaimedTask(t *testing.T) {
+	fake := newFakeCoordinator(t, validClaimedTaskPayload())
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), 0))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if !outcome.Claimed || !outcome.Completed {
+		t.Fatalf("outcome = %+v", outcome)
+	}
+	if len(fake.submits) != 1 {
+		t.Fatalf("submits = %d", len(fake.submits))
+	}
+	result := fake.submits[0]["result"].(map[string]any)
+	if result["artifact_id"] != "33333333-3333-4333-8333-333333333333" {
+		t.Errorf("result artifact = %v", result)
+	}
+	metrics := fake.submits[0]["metrics"].(map[string]any)
+	if metrics["rows"] != float64(1) {
+		t.Errorf("metrics = %v", metrics)
+	}
+	if _, ok := metrics["elapsed_seconds"].(float64); !ok {
+		t.Errorf("missing elapsed_seconds: %v", metrics)
+	}
+	if fake.heartbeats < 1 {
+		t.Error("expected at least one heartbeat")
+	}
+	if len(fake.failures) != 0 {
+		t.Errorf("unexpected failures: %v", fake.failures)
+	}
+}
+
+func TestDaemonReportsChecksumMismatchAsPermanentFailure(t *testing.T) {
+	payload := validClaimedTaskPayload()
+	payload["input"].(map[string]any)["sha256"] = strings.Repeat("b", 64)
+	fake := newFakeCoordinator(t, payload)
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), 0))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if outcome.Completed {
+		t.Fatal("task must not complete on checksum mismatch")
+	}
+	if len(fake.failures) != 1 {
+		t.Fatalf("failures = %d", len(fake.failures))
+	}
+	failure := fake.failures[0]
+	if failure["error_code"] != "ValueError" || failure["retryable"] != false {
+		t.Errorf("failure = %v", failure)
+	}
+	if !strings.Contains(failure["error_message"].(string), "checksum") {
+		t.Errorf("message = %v", failure["error_message"])
+	}
+	if len(fake.submits) != 0 {
+		t.Error("no submission expected")
+	}
+}
+
+func TestDaemonReportsPermanentRunnerFailure(t *testing.T) {
+	fake := newFakeCoordinator(t, validClaimedTaskPayload())
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), ExitPermanent))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if outcome.Completed {
+		t.Fatal("task must not complete")
+	}
+	if len(fake.failures) != 1 || fake.failures[0]["retryable"] != false {
+		t.Fatalf("failures = %v", fake.failures)
+	}
+}
+
+func TestDaemonReportsRetryableRunnerFailure(t *testing.T) {
+	fake := newFakeCoordinator(t, validClaimedTaskPayload())
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), 1))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if outcome.Completed {
+		t.Fatal("task must not complete")
+	}
+	if len(fake.failures) != 1 || fake.failures[0]["retryable"] != true {
+		t.Fatalf("failures = %v", fake.failures)
+	}
+}
+
+func TestDaemonLeaseConflictStopsWithoutFailureReport(t *testing.T) {
+	fake := newFakeCoordinator(t, validClaimedTaskPayload())
+	fake.conflict = true
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), 0))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if !outcome.Claimed {
+		t.Fatal("task was claimed")
+	}
+	if len(fake.failures) != 0 {
+		t.Errorf("no failure report expected after lease loss: %v", fake.failures)
+	}
+	if len(fake.submits) != 0 {
+		t.Errorf("no submission expected after lease loss: %v", fake.submits)
+	}
+}
+
+func TestDaemonIdleClaimIsNotCompleted(t *testing.T) {
+	fake := newFakeCoordinator(t, nil)
+	defer fake.close()
+	daemon := testDaemon(t, fake, fakeRunnerScript(t, t.TempDir(), 0))
+
+	outcome, err := daemon.runOnce()
+	if err != nil {
+		t.Fatalf("runOnce: %v", err)
+	}
+	if outcome.Claimed || outcome.Completed {
+		t.Fatalf("outcome = %+v", outcome)
+	}
+}
@@ -0,0 +1,205 @@
+// Package agent implements a Go worker agent: a coordinator client and
+// task-lifecycle supervisor that executes SDK workloads in a Python
+// subprocess. It mirrors the Python worker's v1 wire contract exactly; the
+// Python worker remains the reference implementation.
+package agent
+
+import (
+	"encoding/json"
+	"fmt"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+)
+
+var (
+	uuidPattern     = regexp.MustCompile(`^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}$`)
+	sha256Pattern   = regexp.MustCompile(`^[0-9a-fA-F]{64}$`)
+	workloadPattern = regexp.MustCompile(`^[a-z][a-z0-9]*(?:[_-][a-z0-9]+)*$`)
+)
+
+// RegisteredWorker is the coordinator's answer to /workers/register.
+type RegisteredWorker struct {
+	WorkerID                 string
+	HeartbeatIntervalSeconds float64
+}
+
+// Input is the claimed task's input artifact.
+type Input struct {
+	URI    string
+	SHA256 string
+}
+
+// Task is one claimed, leased task.
+type Task struct {
+	TaskID          string
+	Attempt         int
+	LeaseExpiresAt  time.Time
+	Workload        string
+	Input           Input
+	Parameters      map[string]any
+	leaseExpiresRaw string
+}
+
+// Uploaded is the coordinator-owned metadata returned after artifact upload.
+type Uploaded struct {
+	ArtifactID string
+	URI        string
+	SHA256     string
+	SizeBytes  int64
+}
+
+func requireString(value any, field string) (string, error) {
+	text, ok := value.(string)
+	if !ok || strings.TrimSpace(text) == "" {
+		return "", fmt.Errorf("%s must be a non-empty string", field)
+	}
+	return text, nil
+}
+
+func safeCoordinatorURI(value any, field string) (string, error) {
+	uri, err := requireString(value, field)
+	if err != nil {
+		return "", err
+	}
+	if strings.HasPrefix(uri, "/") {
+		// A network-path reference (//host/path) or dot segments would
+		// resolve to another origin; reject both.
+		if strings.HasPrefix(uri, "//") {
+			return "", fmt.Errorf("%s must be a safe coordinator path", field)
+		}
+		for _, segment := range strings.Split(uri, "/") {
+			if segment == ".." {
+				return "", fmt.Errorf("%s must be a safe coordinator path", field)
+			}
+		}
+		return uri, nil
+	}
+	parsed, err := url.Parse(uri)
+	if err != nil || (parsed.Scheme != "http" && parsed.Scheme != "https") || parsed.Host == "" {
+		return "", fmt.Errorf("%s must be an absolute HTTP(S) URL or coordinator path", field)
+	}
+	return uri, nil
+}
+
+func sha256Hex(value any, field string) (string, error) {
+	digest, err := requireString(value, field)
+	if err != nil {
+		return "", err
+	}
+	digest = strings.ToLower(digest)
+	if !sha256Pattern.MatchString(digest) {
+		return "", fmt.Errorf("%s must be a SHA-256 hex digest", field)
+	}
+	return digest, nil
+}
+
+func uuid(value any, field string) (string, error) {
+	text, err := requireString(value, field)
+	if err != nil {
+		return "", err
+	}
+	if !uuidPattern.MatchString(text) {
+		return "", fmt.Errorf("%s must be a UUID", field)
+	}
+	return strings.ToLower(text), nil
+}
+
+// ParseTask validates a claimed-task response with the same strictness as the
+// Python worker's ClaimedTask.from_json.
+func ParseTask(payload map[string]any) (*Task, error) {
+	rawInput, ok := payload["input"].(map[string]any)
+	if !ok {
+		return nil, fmt.Errorf("input must be an object")
+	}
+	rawAttempt, ok := payload["attempt"].(float64)
+	if !ok || rawAttempt < 1 || rawAttempt != float64(int(rawAttempt)) {
+		return nil, fmt.Errorf("attempt must be a positive integer")
+	}
+	taskID, err := uuid(payload["task_id"], "task_id")
+	if err != nil {
+		return nil, fmt.Errorf("invalid claimed-task response: %w", err)
+	}
+	rawLease, err := requireString(payload["lease_expires_at"], "lease_expires_at")
+	if err != nil {
+		return nil, fmt.Errorf("invalid claimed-task response: %w", err)
+	}
+	lease, err := time.Parse(time.RFC3339, rawLease)
+	if err != nil || lease.Location() == nil {
+		return nil, fmt.Errorf("lease_expires_at must include a timezone")
+	}
+	workload, err := requireString(payload["workload"], "workload")
+	if err != nil || !workloadPattern.MatchString(workload) {
+		return nil, fmt.Errorf("workload must be a canonical name")
+	}
+	uri, err := safeCoordinatorURI(rawInput["uri"], "input.uri")
+	if err != nil {
+		return nil, fmt.Errorf("invalid claimed-task response: %w", err)
+	}
+	digest, err := sha256Hex(rawInput["sha256"], "input.sha256")
+	if err != nil {
+		return nil, fmt.Errorf("invalid claimed-task response: %w", err)
+	}
+	parameters, ok := payload["parameters"].(map[string]any)
+	if !ok {
+		parameters = map[string]any{}
+	}
+	return &Task{
+		TaskID:          taskID,
+		Attempt:         int(rawAttempt),
+		LeaseExpiresAt:  lease,
+		leaseExpiresRaw: rawLease,
+		Workload:        workload,
+		Input:           Input{URI: uri, SHA256: digest},
+		Parameters:      parameters,
+	}, nil
+}
+
+// LeaseExpiresRaw returns the original lease timestamp string for
+// round-tripping in heartbeat deadlines.
+func (t *Task) LeaseExpiresRaw() string { return t.leaseExpiresRaw }
+
+// ParseRegistered validates a registration response.
+func ParseRegistered(payload map[string]any) (*RegisteredWorker, error) {
+	workerID, err := uuid(payload["worker_id"], "worker_id")
+	if err != nil {
+		return nil, fmt.Errorf("invalid worker registration response: %w", err)
+	}
+	interval, ok := payload["heartbeat_interval_seconds"].(float64)
+	if !ok || interval <= 0 {
+		return nil, fmt.Errorf("heartbeat_interval_seconds must be positive")
+	}
+	return &RegisteredWorker{WorkerID: workerID, HeartbeatIntervalSeconds: interval}, nil
+}
+
+// ParseUploaded validates an artifact upload response.
+func ParseUploaded(payload map[string]any) (*Uploaded, error) {
+	artifactID, err := uuid(payload["artifact_id"], "artifact_id")
+	if err != nil {
+		return nil, fmt.Errorf("invalid artifact upload response: %w", err)
+	}
+	uri, err := safeCoordinatorURI(payload["uri"], "uri")
+	if err != nil {
+		return nil, fmt.Errorf("invalid artifact upload response: %w", err)
+	}
+	digest, err := sha256Hex(payload["sha256"], "sha256")
+	if err != nil {
+		return nil, fmt.Errorf("invalid artifact upload response: %w", err)
+	}
+	rawSize, ok := payload["size_bytes"].(float64)
+	if !ok || rawSize < 0 || rawSize != float64(int64(rawSize)) {
+		return nil, fmt.Errorf("artifact size_bytes must be a non-negative integer")
+	}
+	return &Uploaded{ArtifactID: artifactID, URI: uri, SHA256: digest, SizeBytes: int64(rawSize)}, nil
+}
+
+// TaskRunnerManifest is what the Python task entry writes on success.
+type TaskRunnerManifest struct {
+	ArtifactPath string         `json:"artifact_path"`
+	ContentType  string         `json:"content_type"`
+	Metrics      map[string]any `json:"metrics"`
+}
+
+// Encode serializes a claim payload for /tasks/claim.
+func Encode(v any) ([]byte, error) { return json.Marshal(v) }
@@ -0,0 +1,111 @@
+package agent
+
+import (
+	"testing"
+	"time"
+)
+
+func validTaskPayload() map[string]any {
+	return map[string]any{
+		"task_id":          "11111111-1111-4111-8111-111111111111",
+		"attempt":          1.0,
+		"lease_expires_at": "2026-08-02T00:00:00Z",
+		"workload":         "similarity-search",
+		"input": map[string]any{
+			"uri":    "/tasks/11111111-1111-4111-8111-111111111111/input",
+			"sha256": "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
+		},
+		"parameters": map[string]any{"query_smiles": "CCO"},
+	}
+}
+
+func TestParseTaskAcceptsValidPayload(t *testing.T) {
+	task, err := ParseTask(validTaskPayload())
+	if err != nil {
+		t.Fatalf("ParseTask: %v", err)
+	}
+	if task.TaskID != "11111111-1111-4111-8111-111111111111" {
+		t.Errorf("task id = %q", task.TaskID)
+	}
+	if task.Attempt != 1 || task.Workload != "similarity-search" {
+		t.Errorf("attempt/workload = %d/%q", task.Attempt, task.Workload)
+	}
+	if task.Input.SHA256 != "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" {
+		t.Errorf("sha256 = %q", task.Input.SHA256)
+	}
+	if task.LeaseExpiresAt.IsZero() {
+		t.Error("lease must parse")
+	}
+}
+
+func TestParseTaskRejectsInvalidPayloads(t *testing.T) {
+	tests := []struct {
+		name   string
+		mutate func(map[string]any)
+	}{
+		{"non-uuid task id", func(p map[string]any) { p["task_id"] = "../outside" }},
+		{"zero attempt", func(p map[string]any) { p["attempt"] = 0 }},
+		{"naive lease", func(p map[string]any) { p["lease_expires_at"] = "2026-08-02T00:00:00" }},
+		{"network-path uri", func(p map[string]any) {
+			p["input"].(map[string]any)["uri"] = "//outside.example/input"
+		}},
+		{"dot-segment uri", func(p map[string]any) {
+			p["input"].(map[string]any)["uri"] = "/tasks/../outside/input"
+		}},
+		{"short sha256", func(p map[string]any) {
+			p["input"].(map[string]any)["sha256"] = "abc"
+		}},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			payload := validTaskPayload()
+			test.mutate(payload)
+			if _, err := ParseTask(payload); err == nil {
+				t.Error("expected ParseTask to reject the payload")
+			}
+		})
+	}
+}
+
+func TestParseRegisteredAndUploaded(t *testing.T) {
+	registered, err := ParseRegistered(map[string]any{
+		"worker_id":                  "22222222-2222-4222-8222-222222222222",
+		"heartbeat_interval_seconds": 15.0,
+	})
+	if err != nil {
+		t.Fatalf("ParseRegistered: %v", err)
+	}
+	if registered.HeartbeatIntervalSeconds != 15 {
+		t.Errorf("interval = %v", registered.HeartbeatIntervalSeconds)
+	}
+
+	uploaded, err := ParseUploaded(map[string]any{
+		"artifact_id": "33333333-3333-4333-8333-333333333333",
+		"uri":         "https://coordinator.example/artifacts/333/download",
+		"sha256":      "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb",
+		"size_bytes":  12.0,
+	})
+	if err != nil {
+		t.Fatalf("ParseUploaded: %v", err)
+	}
+	if uploaded.SizeBytes != 12 {
+		t.Errorf("size = %d", uploaded.SizeBytes)
+	}
+
+	if _, err := ParseUploaded(map[string]any{"artifact_id": "missing"}); err == nil {
+		t.Error("expected invalid upload metadata to fail")
+	}
+}
+
+func TestLeaseHeartbeatDelayIsBelowHalfTTL(t *testing.T) {
+	task, err := ParseTask(validTaskPayload())
+	if err != nil {
+		t.Fatal(err)
+	}
+	task.LeaseExpiresAt = time.Now().Add(60 * time.Second)
+	heartbeat := newLeaseHeartbeat(task, "worker", nil, 15*time.Second)
+	delay := heartbeat.nextDelay()
+	if delay > 30*time.Second || delay <= 0 {
+		t.Errorf("delay = %v, want < 30s", delay)
+	}
+}
@@ -0,0 +1,73 @@
+package agent
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net/http"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+)
+
+// ReleaseWheelURL returns the download URL of the scimesh wheel attached to
+// the GitHub release that matches the given binary version (for example
+// "1.1.0-alpha.10"), plus the wheel file name. The wheel is version-locked to
+// the binary so a worker's catalog always matches its task runner.
+func ReleaseWheelURL(version string) (string, string, error) {
+	if version == "" || version == "dev" {
+		return "", "", fmt.Errorf("no release wheel for build %q", version)
+	}
+	filename := fmt.Sprintf("scimesh-%s-py3-none-any.whl", NormalizePEP440(version))
+	return fmt.Sprintf("https://github.com/emil28092005/SciMesh/releases/download/v%s/%s", version, filename), filename, nil
+}
+
+// NormalizePEP440 turns our release tag suffixes into the PEP 440 form
+// setuptools uses for wheel names: 1.1.0-alpha.10 -> 1.1.0a10,
+// 1.1.0-beta.2 -> 1.1.0b2, 1.1.0-rc.1 -> 1.1.0rc1. Stable tags pass through.
+func NormalizePEP440(version string) string {
+	for from, to := range map[string]string{"-alpha.": "a", "-beta.": "b", "-rc.": "rc"} {
+		version = strings.ReplaceAll(version, from, to)
+	}
+	return version
+}
+
+// DownloadWheel fetches the release wheel into dir (config directory of the
+// wizard / serve data dir) and returns the local path. Best-effort download
+// with a generous timeout: wheels can be several MB.
+func DownloadWheel(ctx context.Context, url, dir string) (string, error) {
+	target := filepath.Join(dir, wheelNameFromURL(url))
+	if err := os.MkdirAll(dir, 0o700); err != nil {
+		return "", fmt.Errorf("download wheel: %w", err)
+	}
+	ctx, cancel := context.WithTimeout(ctx, 10*time.Minute)
+	defer cancel()
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
+	if err != nil {
+		return "", err
+	}
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("download wheel: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("download wheel: HTTP %d", resp.StatusCode)
+	}
+	//nolint:gosec // G304: target is our own config dir + a fixed wheel name
+	out, err := os.Create(target)
+	if err != nil {
+		return "", fmt.Errorf("download wheel: %w", err)
+	}
+	defer func() { _ = out.Close() }()
+	if _, err := io.Copy(out, resp.Body); err != nil {
+		return "", fmt.Errorf("download wheel: %w", err)
+	}
+	return target, nil
+}
+
+// wheelNameFromURL extracts the trailing file name of a wheel URL.
+func wheelNameFromURL(url string) string {
+	return url[strings.LastIndex(url, "/")+1:]
+}
@@ -0,0 +1,81 @@
+package agent
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"strings"
+	"testing"
+)
+
+func TestNormalizePEP440(t *testing.T) {
+	cases := map[string]string{
+		"1.1.0":          "1.1.0",
+		"1.1.0-alpha.10": "1.1.0a10",
+		"1.1.0-beta.2":   "1.1.0b2",
+		"1.1.0-rc.1":     "1.1.0rc1",
+		"1.0.0":          "1.0.0",
+	}
+	for in, want := range cases {
+		if got := NormalizePEP440(in); got != want {
+			t.Errorf("NormalizePEP440(%q) = %q, want %q", in, got, want)
+		}
+	}
+}
+
+func TestReleaseWheelURL(t *testing.T) {
+	url, name, err := ReleaseWheelURL("1.1.0-alpha.10")
+	if err != nil {
+		t.Fatal(err)
+	}
+	wantURL := "https://github.com/emil28092005/SciMesh/releases/download/v1.1.0-alpha.10/scimesh-1.1.0a10-py3-none-any.whl"
+	if url != wantURL {
+		t.Errorf("url = %q, want %q", url, wantURL)
+	}
+	if name != "scimesh-1.1.0a10-py3-none-any.whl" {
+		t.Errorf("name = %q", name)
+	}
+
+	// A dev build has no release wheel.
+	if _, _, err := ReleaseWheelURL("dev"); err == nil {
+		t.Error("dev build must not resolve a release wheel")
+	}
+	if _, _, err := ReleaseWheelURL(""); err == nil {
+		t.Error("empty version must not resolve a release wheel")
+	}
+}
+
+func TestDownloadWheel(t *testing.T) {
+	payload := []byte("fake wheel bytes")
+	stub := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		_, _ = w.Write(payload)
+	}))
+	defer stub.Close()
+
+	dir := t.TempDir()
+	path, err := DownloadWheel(context.Background(), stub.URL+"/scimesh-1.1.0a10-py3-none-any.whl", dir)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !strings.HasSuffix(path, "scimesh-1.1.0a10-py3-none-any.whl") {
+		t.Errorf("path = %q", path)
+	}
+	got, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(got) != string(payload) {
+		t.Error("wheel bytes mismatch")
+	}
+}
+
+func TestDownloadWheelReportsHTTPErrors(t *testing.T) {
+	stub := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		http.NotFound(w, r)
+	}))
+	defer stub.Close()
+	if _, err := DownloadWheel(context.Background(), stub.URL+"/missing.whl", t.TempDir()); err == nil {
+		t.Error("404 must fail the download")
+	}
+}
@@ -0,0 +1,58 @@
+package agent
+
+import (
+	"errors"
+	"fmt"
+	"os"
+	"regexp"
+	"strings"
+)
+
+var (
+	// Go's regexp (RE2) has no lookbehind, so these patterns conservatively
+	// anchor on the characters that typically precede a local path:
+	// whitespace, quotes, parens, brackets, or the start of the message.
+	windowsPathPattern = regexp.MustCompile(`[A-Za-z]:\\[^\s'"\],)]+`)
+	posixPathPattern   = regexp.MustCompile(`(^|[\s'"(\[=])/(?:[^\s'"\],)]+)`)
+)
+
+// SanitizeErrorMessage keeps coordinator-visible failures useful without
+// exposing local paths. It mirrors the Python worker's sanitizer: local work
+// directories and absolute paths are redacted, and the message is truncated
+// to 300 characters.
+func SanitizeErrorMessage(message string, workDir string) string {
+	message = strings.ReplaceAll(message, workDir, "<worker-dir>")
+	message = windowsPathPattern.ReplaceAllString(message, "<path>")
+	message = posixPathPattern.ReplaceAllString(message, "${1}<path>")
+	if len(message) > 300 {
+		message = message[:300]
+	}
+	return message
+}
+
+// IsRetryableError classifies failures for the coordinator. Invalid scientific
+// input and missing local tools are permanent; everything else (transient
+// transport errors, unexpected runner failures) may be retried.
+func IsRetryableError(err error) bool {
+	if err == nil {
+		return false
+	}
+	var coordinatorErr *CoordinatorError
+	var pathErr *os.PathError
+	if errors.As(err, &coordinatorErr) || errors.As(err, &pathErr) {
+		return false
+	}
+	return true
+}
+
+// TaskRunnerExit classifies subprocess exits.
+const (
+	ExitPermanent = 3 // runner classified the failure as invalid input
+)
+
+func runnerExitError(exit int, stderr string) error {
+	if exit == ExitPermanent {
+		return &CoordinatorError{msg: stderr}
+	}
+	return fmt.Errorf("task runner failed with exit code %d: %s", exit, stderr)
+}
@@ -0,0 +1,625 @@
+// Package setupui serves the local worker setup wizard: a small HTTP server
+// bound to 127.0.0.1 that writes the worker's config file, runs preflight
+// checks, and starts/stops the worker as a background process. It is part of
+// the worker-agent binary so a machine that installs only a worker never needs
+// a coordinator.
+package setupui
+
+import (
+	"context"
+	"embed"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"log/slog"
+	"net"
+	"net/http"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent"
+)
+
+//go:embed template.html
+var templateFS embed.FS
+
+const (
+	defaultPort = 12700
+	pidFileName = "worker.pid"
+	logFileName = "worker.log"
+)
+
+// Supervisor starts and stops the worker process and tracks its pid. It is an
+// interface so tests can substitute a fake.
+type Supervisor interface {
+	// Start launches `worker-agent --config <path>` detached, writing output
+	// into the log file. Returns the child pid.
+	Start(configPath, logPath string) (int, error)
+	// Stop terminates the process recorded in the pid file.
+	Stop() error
+	// Pid returns the recorded child pid, or 0 when none is recorded.
+	Pid() int
+	// Alive reports whether the recorded child is still running.
+	Alive() bool
+}
+
+// PIDSupervisor is the real Supervisor: it spawns the running binary with
+// --config and manages its pid file. Liveness comes from a Wait goroutine, so
+// it works on every platform (no signal probing, which Windows lacks).
+type PIDSupervisor struct {
+	mu      sync.Mutex
+	pidPath string
+	proc    *os.Process
+	done    chan struct{} // closed when the spawned process exits; nil when not started
+}
+
+func NewPIDSupervisor(pidPath string) *PIDSupervisor { return &PIDSupervisor{pidPath: pidPath} }
+
+func (s *PIDSupervisor) Pid() int {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return s.readPid()
+}
+
+func (s *PIDSupervisor) readPid() int {
+	raw, err := os.ReadFile(s.pidPath)
+	if err != nil {
+		return 0
+	}
+	pid, err := strconv.Atoi(strings.TrimSpace(string(raw)))
+	if err != nil || pid < 1 {
+		return 0
+	}
+	return pid
+}
+
+func (s *PIDSupervisor) Alive() bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if s.done == nil {
+		return false
+	}
+	select {
+	case <-s.done:
+		return false
+	default:
+		return true
+	}
+}
+
+func (s *PIDSupervisor) Start(configPath, logPath string) (int, error) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if s.done != nil {
+		select {
+		case <-s.done:
+		default:
+			return s.readPid(), fmt.Errorf("worker is already running (pid %d)", s.readPid())
+		}
+	}
+	exe, err := os.Executable()
+	if err != nil {
+		return 0, fmt.Errorf("resolve worker binary: %w", err)
+	}
+	//nolint:gosec // G304: logPath lives in the wizard's own config directory
+	logFile, err := os.OpenFile(logPath, os.O_CREATE|os.O_APPEND|os.O_WRONLY, 0o600)
+	if err != nil {
+		return 0, fmt.Errorf("open worker log: %w", err)
+	}
+	defer func() { _ = logFile.Close() }()
+	//nolint:gosec // G204: exe is os.Executable, configPath is the wizard's own file;
+	// Background ctx: the child's lifecycle is managed by the supervisor, not the context
+	cmd := exec.CommandContext(context.Background(), exe, "--config", configPath)
+	cmd.Stdout = logFile
+	cmd.Stderr = logFile
+	cmd.Stdin = nil
+	if err := cmd.Start(); err != nil {
+		return 0, fmt.Errorf("start worker: %w", err)
+	}
+	// The child inherits our stdout/stderr descriptors pointing at the log
+	// file, so we can close our copy; the child keeps it open.
+	_ = logFile.Close()
+	s.proc = cmd.Process
+	s.done = make(chan struct{})
+	go func() { _ = cmd.Wait(); close(s.done) }()
+	if err := os.WriteFile(s.pidPath, []byte(strconv.Itoa(cmd.Process.Pid)+"\n"), 0o600); err != nil {
+		_ = cmd.Process.Kill()
+		return 0, fmt.Errorf("write pid file: %w", err)
+	}
+	return cmd.Process.Pid, nil
+}
+
+func (s *PIDSupervisor) Stop() error {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if s.done == nil {
+		return nil
+	}
+	select {
+	case <-s.done:
+		s.done = nil
+		s.proc = nil
+		_ = os.Remove(s.pidPath)
+		return nil
+	default:
+	}
+	// Ask politely, then force. os.Interrupt terminates on Windows too.
+	_ = s.proc.Signal(os.Interrupt)
+	deadline := time.Now().Add(3 * time.Second)
+	for time.Now().Before(deadline) {
+		select {
+		case <-s.done:
+			s.done = nil
+			s.proc = nil
+			_ = os.Remove(s.pidPath)
+			return nil
+		default:
+			time.Sleep(100 * time.Millisecond)
+		}
+	}
+	_ = s.proc.Kill()
+	select {
+	case <-s.done:
+	case <-time.After(2 * time.Second):
+	}
+	s.done = nil
+	s.proc = nil
+	_ = os.Remove(s.pidPath)
+	return nil
+}
+
+// Server is the wizard HTTP server, bound to the loopback interface only.
+type Server struct {
+	log           *slog.Logger
+	cfgPath       string
+	logPath       string
+	dir           string
+	sup           Supervisor
+	openBrowser   func(string)
+	port          int
+	install       func(ctx context.Context, venvPython, pkg string) error
+	downloadWheel func(ctx context.Context, url, dir string) (string, error)
+}
+
+// Options customises the wizard for tests and embedding.
+type Options struct {
+	Port        int
+	ConfigPath  string
+	OpenBrowser func(url string)
+	Supervisor  Supervisor
+	Dir         string // directory for pid/log files; defaults to the config dir
+	// InstallScimesh overrides the pip step of the runtime installer (tests
+	// substitute a fake); nil uses the real pip inside the managed venv.
+	InstallScimesh func(ctx context.Context, venvPython, pkg string) error
+	// DownloadWheel overrides the release-wheel download (tests substitute a
+	// fake); nil downloads from the GitHub release matching the agent version.
+	DownloadWheel func(ctx context.Context, url, dir string) (string, error)
+}
+
+func New(log *slog.Logger, opts Options) *Server {
+	cfgPath := opts.ConfigPath
+	if cfgPath == "" {
+		cfgPath = agent.DefaultConfigPath()
+	}
+	dir := opts.Dir
+	if dir == "" {
+		dir = filepath.Dir(cfgPath)
+	}
+	sup := opts.Supervisor
+	if sup == nil {
+		sup = NewPIDSupervisor(filepath.Join(dir, pidFileName))
+	}
+	open := opts.OpenBrowser
+	if open == nil {
+		open = func(string) {}
+	}
+	port := opts.Port
+	if port == 0 {
+		port = defaultPort
+	}
+	install := opts.InstallScimesh
+	if install == nil {
+		install = installScimeshWithPip
+	}
+	downloadWheel := opts.DownloadWheel
+	if downloadWheel == nil {
+		downloadWheel = agent.DownloadWheel
+	}
+	return &Server{log: log, cfgPath: cfgPath, logPath: filepath.Join(dir, logFileName), dir: dir, sup: sup, openBrowser: open, port: port, install: install, downloadWheel: downloadWheel}
+}
+
+// Listen binds the loopback listener and returns it; Serve runs the server on
+// it. Split so tests can inspect the actual ephemeral port.
+func (s *Server) Listen() (net.Listener, error) {
+	return (&net.ListenConfig{}).Listen(context.Background(), "tcp", fmt.Sprintf("127.0.0.1:%d", s.port))
+}
+
+// OpenBrowser hands the wizard URL to the configured opener (default: no-op).
+func (s *Server) OpenBrowser(url string) { s.openBrowser(url) }
+
+// Serve runs the wizard until ctx is cancelled.
+func (s *Server) Serve(ctx context.Context, listener net.Listener) error {
+	mux := http.NewServeMux()
+	mux.HandleFunc("GET /", s.handleIndex)
+	mux.HandleFunc("GET /api/status", s.handleStatus)
+	mux.HandleFunc("POST /api/config", s.handleSaveConfig)
+	mux.HandleFunc("POST /api/test", s.handleTest)
+	mux.HandleFunc("POST /api/runtime/install", s.handleInstallRuntime)
+	mux.HandleFunc("POST /api/start", s.handleStart)
+	mux.HandleFunc("POST /api/stop", s.handleStop)
+	mux.HandleFunc("GET /api/logs", s.handleLogs)
+	server := &http.Server{
+		Handler:           mux,
+		ReadHeaderTimeout: 10 * time.Second,
+	}
+	go func() {
+		<-ctx.Done()
+		shutdownCtx, cancel := context.WithTimeout(context.WithoutCancel(ctx), 3*time.Second)
+		defer cancel()
+		_ = server.Shutdown(shutdownCtx)
+	}()
+	return server.Serve(listener)
+}
+
+func (s *Server) handleIndex(w http.ResponseWriter, r *http.Request) {
+	if r.URL.Path != "/" {
+		http.NotFound(w, r)
+		return
+	}
+	html, err := templateFS.ReadFile("template.html")
+	if err != nil {
+		http.Error(w, "template unavailable", http.StatusInternalServerError)
+		return
+	}
+	w.Header().Set("Content-Type", "text/html; charset=utf-8")
+	_, _ = w.Write(html)
+}
+
+func writeJSON(w http.ResponseWriter, status int, v any) {
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(status)
+	_ = json.NewEncoder(w).Encode(v)
+}
+
+// statusView is what the wizard needs to paint the running/stopped state.
+type statusView struct {
+	ConfigPresent bool        `json:"config_present"`
+	ConfigPath    string      `json:"config_path"`
+	LogPath       string      `json:"log_path"`
+	Running       bool        `json:"running"`
+	Pid           int         `json:"pid"`
+	WorkerName    string      `json:"worker_name,omitempty"`
+	Coordinator   string      `json:"coordinator,omitempty"`
+	WorkDir       string      `json:"work_dir,omitempty"`
+	TokenSet      bool        `json:"token_set"`
+	Stats         WorkerStats `json:"stats"`
+}
+
+// WorkerStats is parsed from the worker log: the agent reports each claim,
+// completion and failure as a structured line, so the wizard can show live
+// counters without any coordinator access.
+type WorkerStats struct {
+	Registered bool `json:"registered"`
+	Claimed    int  `json:"claimed"`
+	Completed  int  `json:"completed"`
+	Failed     int  `json:"failed"`
+}
+
+// ensureVenvTaskRunner rewrites the saved config so its task runner uses the
+// managed venv python when one exists and the config does not already pin one.
+func (s *Server) ensureVenvTaskRunner() {
+	raw, err := os.ReadFile(s.cfgPath)
+	if err != nil {
+		return
+	}
+	var file agent.ConfigFile
+	if json.Unmarshal(raw, &file) != nil || len(file.TaskRunner) > 0 {
+		return
+	}
+	if venv := s.venvPython(); venv != "" {
+		file.TaskRunner = []string{venv, "-m", "scimesh.worker.task"}
+		if payload, err := json.MarshalIndent(file, "", "  "); err == nil {
+			_ = os.WriteFile(s.cfgPath, append(payload, '\n'), 0o600)
+		}
+	}
+}
+
+// parseWorkerStats counts the structured agent events in the worker log.
+func parseWorkerStats(logPath string) WorkerStats {
+	var stats WorkerStats
+	//nolint:gosec // G304: logPath is the wizard's own log file next to the config
+	raw, err := os.ReadFile(logPath)
+	if err != nil {
+		return stats
+	}
+	for _, line := range strings.Split(string(raw), "\n") {
+		switch {
+		case strings.Contains(line, "msg=registered"):
+			stats.Registered = true
+		case strings.Contains(line, `msg="task claimed"`):
+			stats.Claimed++
+		case strings.Contains(line, `msg="task completed"`):
+			stats.Completed++
+		case strings.Contains(line, `msg="task failed"`):
+			stats.Failed++
+		}
+	}
+	return stats
+}
+
+func (s *Server) handleStatus(w http.ResponseWriter, r *http.Request) {
+	view := statusView{ConfigPath: s.cfgPath, LogPath: s.logPath, Running: s.sup.Alive(), Pid: s.sup.Pid(), Stats: parseWorkerStats(s.logPath)}
+	if raw, err := os.ReadFile(s.cfgPath); err == nil {
+		var file agent.ConfigFile
+		if json.Unmarshal(raw, &file) == nil {
+			view.ConfigPresent = true
+			view.WorkerName = file.WorkerName
+			view.Coordinator = file.CoordinatorURL
+			view.WorkDir = file.WorkDir
+			view.TokenSet = file.Token != "" || file.WorkerKey != ""
+		}
+	}
+	writeJSON(w, http.StatusOK, view)
+}
+
+type saveConfigRequest struct {
+	CoordinatorURL string   `json:"coordinator_url"`
+	Token          string   `json:"token"`
+	WorkerKey      string   `json:"worker_key"`
+	UserserviceURL string   `json:"userservice_url"`
+	WorkDir        string   `json:"work_dir"`
+	WorkerName     string   `json:"worker_name"`
+	CPUCount       int      `json:"cpu_count"`
+	MemoryMB       int      `json:"memory_mb"`
+	TaskRunner     []string `json:"task_runner"`
+}
+
+func (s *Server) handleSaveConfig(w http.ResponseWriter, r *http.Request) {
+	var req saveConfigRequest
+	if err := json.NewDecoder(io.LimitReader(r.Body, 1<<20)).Decode(&req); err != nil {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "invalid JSON body"})
+		return
+	}
+	file := agent.ConfigFile{
+		CoordinatorURL: strings.TrimSpace(req.CoordinatorURL),
+		Token:          req.Token,
+		WorkerKey:      req.WorkerKey,
+		UserserviceURL: strings.TrimSpace(req.UserserviceURL),
+		WorkDir:        strings.TrimSpace(req.WorkDir),
+		WorkerName:     strings.TrimSpace(req.WorkerName),
+		CPUCount:       req.CPUCount,
+		MemoryMB:       req.MemoryMB,
+		TaskRunner:     req.TaskRunner,
+	}
+	if file.CoordinatorURL == "" {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "coordinator_url is required"})
+		return
+	}
+	if !strings.HasPrefix(file.CoordinatorURL, "http://") && !strings.HasPrefix(file.CoordinatorURL, "https://") {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "coordinator_url must be an absolute HTTP(S) URL"})
+		return
+	}
+	if file.WorkDir == "" {
+		file.WorkDir = "./scimesh-agent-data"
+	}
+	if file.WorkerName == "" {
+		if host, err := os.Hostname(); err == nil {
+			file.WorkerName = host
+		} else {
+			file.WorkerName = "worker"
+		}
+	}
+	if file.CPUCount < 1 {
+		file.CPUCount = 1
+	}
+	// The wizard UI bakes the venv python into the runner after an install;
+	// an API-driven or scripted flow may not, so the server guarantees it:
+	// workloads execute through scimesh's task runner, which lives in the venv.
+	if len(file.TaskRunner) == 0 {
+		if venv := s.venvPython(); venv != "" {
+			file.TaskRunner = []string{venv, "-m", "scimesh.worker.task"}
+		}
+	}
+	if err := agent.SaveConfigFile(s.cfgPath, file); err != nil {
+		s.log.Error("save wizard config", "err", err)
+		writeJSON(w, http.StatusInternalServerError, map[string]string{"error": "could not write the config file"})
+		return
+	}
+	writeJSON(w, http.StatusOK, map[string]bool{"saved": true})
+}
+
+func (s *Server) handleTest(w http.ResponseWriter, r *http.Request) {
+	var req saveConfigRequest
+	if err := json.NewDecoder(io.LimitReader(r.Body, 1<<20)).Decode(&req); err != nil {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "invalid JSON body"})
+		return
+	}
+	url := strings.TrimSpace(req.CoordinatorURL)
+	if url == "" {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "coordinator_url is required"})
+		return
+	}
+	// After the runtime installer created the venv, probe that interpreter:
+	// checking the bare system python3 would keep reporting scimesh as
+	// missing even though the worker would run with the venv.
+	report := agent.RunCheck(r.Context(), url, s.venvPython(), req.Token, req.WorkerKey, req.UserserviceURL)
+	writeJSON(w, http.StatusOK, report)
+}
+
+func (s *Server) handleStart(w http.ResponseWriter, r *http.Request) {
+	if _, err := os.Stat(s.cfgPath); err != nil {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "no configuration saved yet"})
+		return
+	}
+	s.ensureVenvTaskRunner()
+	pid, err := s.sup.Start(s.cfgPath, s.logPath)
+	if err != nil {
+		writeJSON(w, http.StatusConflict, map[string]string{"error": err.Error()})
+		return
+	}
+	writeJSON(w, http.StatusOK, map[string]int{"pid": pid})
+}
+
+func (s *Server) handleStop(w http.ResponseWriter, r *http.Request) {
+	if err := s.sup.Stop(); err != nil {
+		writeJSON(w, http.StatusConflict, map[string]string{"error": err.Error()})
+		return
+	}
+	writeJSON(w, http.StatusOK, map[string]bool{"stopped": true})
+}
+
+func (s *Server) handleLogs(w http.ResponseWriter, r *http.Request) {
+	raw, err := os.ReadFile(s.logPath)
+	if err != nil {
+		writeJSON(w, http.StatusOK, map[string]string{"log": ""})
+		return
+	}
+	lines := strings.Split(strings.TrimRight(string(raw), "\n"), "\n")
+	tail := 200
+	if n, err := strconv.Atoi(r.URL.Query().Get("tail")); err == nil && n > 0 && n < 5000 {
+		tail = n
+	}
+	if len(lines) > tail {
+		lines = lines[len(lines)-tail:]
+	}
+	writeJSON(w, http.StatusOK, map[string]string{"log": strings.Join(lines, "\n")})
+}
+
+// ErrCanceled mirrors context.Canceled for callers that treat a cancelled
+// wizard as a clean exit.
+var ErrCanceled = errors.New("setup wizard cancelled")
+
+type installRuntimeRequest struct {
+	// ScimeshPackage overrides where the scimesh wheel comes from: a local
+	// wheel/index path or the PyPI name. Defaults to SCIMESH_PIP_PACKAGE, then
+	// to the PyPI name.
+	ScimeshPackage string `json:"scimesh_package"`
+}
+
+type installRuntimeResponse struct {
+	OK        bool   `json:"ok"`
+	Python    string `json:"python,omitempty"` // venv python to use as TASK_RUNNER[0]
+	Installed bool   `json:"installed"`
+}
+
+// handleInstallRuntime creates a managed venv next to the worker config and
+// installs the scimesh package into it, so the machine needs no manual pip
+// step. The venv python path is returned for the wizard to bake into the
+// task runner.
+func (s *Server) handleInstallRuntime(w http.ResponseWriter, r *http.Request) {
+	// The wizard may run the install before any config was saved, so the
+	// worker directory (venv, wheel) may not exist yet.
+	if err := os.MkdirAll(s.dir, 0o700); err != nil {
+		writeJSON(w, http.StatusConflict, map[string]string{"error": "could not create the worker directory"})
+		return
+	}
+	var req installRuntimeRequest
+	if err := json.NewDecoder(io.LimitReader(r.Body, 1<<20)).Decode(&req); err != nil {
+		writeJSON(w, http.StatusBadRequest, map[string]string{"error": "invalid JSON body"})
+		return
+	}
+	pkg := strings.TrimSpace(req.ScimeshPackage)
+	if pkg == "" {
+		pkg = os.Getenv("SCIMESH_PIP_PACKAGE")
+	}
+	if pkg == "" {
+		// No PyPI default on purpose: the PyPI name "scimesh" belongs to an
+		// unrelated project. Instead we ship the wheel in our own GitHub
+		// release, version-locked to this binary, and download it from there.
+		url, _, err := agent.ReleaseWheelURL(agent.Version)
+		if err != nil {
+			writeJSON(w, http.StatusConflict, map[string]string{
+				"error": "no scimesh source configured: set SCIMESH_PIP_PACKAGE to your wheel, checkout or index, then retry",
+			})
+			return
+		}
+		local, err := s.downloadWheel(r.Context(), url, s.dir)
+		if err != nil {
+			s.log.Error("download release wheel", "err", err, "url", url)
+			writeJSON(w, http.StatusConflict, map[string]string{
+				"error": "could not download the scimesh wheel for this release: " + err.Error() + ". Set SCIMESH_PIP_PACKAGE to your wheel, checkout or index and retry.",
+			})
+			return
+		}
+		pkg = local
+	}
+
+	python3, err := exec.LookPath("python3")
+	if err != nil {
+		writeJSON(w, http.StatusConflict, map[string]string{"error": "python3 is not installed on this machine"})
+		return
+	}
+	venvDir := filepath.Join(s.dir, "venv")
+	venvPython := filepath.Join(venvDir, "bin", "python")
+	if _, err := os.Stat(venvPython); err != nil {
+		// Windows layout: Scripts/python.exe.
+		if win := filepath.Join(venvDir, "Scripts", "python.exe"); stat(win) {
+			venvPython = win
+		}
+	}
+	if _, err := os.Stat(venvPython); err != nil {
+		ctx, cancel := context.WithTimeout(r.Context(), 3*time.Minute)
+		defer cancel()
+		cmd := exec.CommandContext(ctx, python3, "-m", "venv", venvDir) //nolint:gosec // G204: python3 from LookPath, venvDir is our own dir
+		if out, err := cmd.CombinedOutput(); err != nil {
+			s.log.Error("create runtime venv", "err", err, "out", truncate(string(out), 500))
+			writeJSON(w, http.StatusConflict, map[string]string{"error": "could not create the python venv"})
+			return
+		}
+	}
+
+	if err := s.install(r.Context(), venvPython, pkg); err != nil {
+		s.log.Error("install scimesh runtime", "err", err)
+		writeJSON(w, http.StatusConflict, map[string]string{
+			"error": "pip install " + pkg + " failed: " + err.Error() +
+				". Set SCIMESH_PIP_PACKAGE to your scimesh wheel or index and retry.",
+		})
+		return
+	}
+	writeJSON(w, http.StatusOK, installRuntimeResponse{OK: true, Python: venvPython, Installed: true})
+}
+
+// venvPython returns the managed venv python when the runtime installer has
+// created one, so the task runner can be pointed at it automatically.
+func (s *Server) venvPython() string {
+	for _, candidate := range []string{
+		filepath.Join(s.dir, "venv", "bin", "python"),
+		filepath.Join(s.dir, "venv", "Scripts", "python.exe"),
+	} {
+		if _, err := os.Stat(candidate); err == nil {
+			return candidate
+		}
+	}
+	return ""
+}
+
+// installScimeshWithPip installs the package with the venv's own pip,
+// streaming into the agent log so a long build is not silent.
+func installScimeshWithPip(ctx context.Context, venvPython, pkg string) error {
+	pip := filepath.Join(filepath.Dir(venvPython), "pip")
+	if _, err := os.Stat(pip); err != nil {
+		pip += ".exe"
+	}
+	ctx, cancel := context.WithTimeout(ctx, 10*time.Minute)
+	defer cancel()
+	cmd := exec.CommandContext(ctx, pip, "install", pkg) //nolint:gosec // G204: pip from our venv, pkg is operator-set or a fixed default
+	return cmd.Run()
+}
+
+func stat(path string) bool {
+	_, err := os.Stat(path)
+	return err == nil
+}
+
+func truncate(s string, n int) string {
+	if len(s) <= n {
+		return s
+	}
+	return s[:n] + "…"
+}
@@ -0,0 +1,499 @@
+package setupui
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"io"
+	"log/slog"
+	"net"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/agent"
+)
+
+func testLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+func newTestServer(t *testing.T, sup Supervisor) (*Server, string) {
+	t.Helper()
+	return newTestServerWithInstall(t, sup, nil)
+}
+
+func newTestServerWithInstall(t *testing.T, sup Supervisor, install func(ctx context.Context, venvPython, pkg string) error) (*Server, string) {
+	t.Helper()
+	return newTestServerWithInstallAndWheel(t, sup, install, nil)
+}
+
+func newTestServerWithInstallAndWheel(t *testing.T, sup Supervisor, install func(ctx context.Context, venvPython, pkg string) error, wheel func(ctx context.Context, url, dir string) (string, error)) (*Server, string) {
+	t.Helper()
+	dir := t.TempDir()
+	server := New(testLogger(), Options{
+		// A distinct random port per test: Port 0 means "the default 12700" in
+		// the server, which would let the shared http.Client pool reuse a stale
+		// keep-alive connection across tests (EOF after a Shutdown).
+		Port:           freePort(t),
+		ConfigPath:     filepath.Join(dir, "config.json"),
+		Dir:            dir,
+		Supervisor:     sup,
+		OpenBrowser:    func(string) {},
+		InstallScimesh: install,
+		DownloadWheel:  wheel,
+	})
+	listener, err := server.Listen()
+	if err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() { _ = listener.Close() })
+	ctx, cancel := context.WithCancel(context.Background())
+	t.Cleanup(cancel)
+	go func() { _ = server.Serve(ctx, listener) }()
+	return server, "http://" + listener.Addr().String()
+}
+
+type fakeSup struct {
+	mu      sync.Mutex
+	started bool
+	stopped bool
+	pid     int
+	alive   bool
+}
+
+func (f *fakeSup) Start(configPath, logPath string) (int, error) {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	f.started = true
+	f.alive = true
+	f.pid = 4242
+	return f.pid, nil
+}
+
+func (f *fakeSup) Stop() error {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	f.stopped = true
+	f.alive = false
+	return nil
+}
+
+func (f *fakeSup) Pid() int { f.mu.Lock(); defer f.mu.Unlock(); return f.pid }
+func (f *fakeSup) Alive() bool {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	return f.alive
+}
+
+func postJSON(t *testing.T, base, path string, body any) (*httptest.ResponseRecorder, map[string]any) {
+	t.Helper()
+	req, err := http.NewRequestWithContext(context.Background(), http.MethodPost, base+path, strings.NewReader(mustJSON(t, body)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	req.Header.Set("Content-Type", "application/json")
+	// No keep-alive pooling: a pooled connection to a shut-down test server
+	// would surface as an EOF instead of a fresh dial.
+	client := http.Client{Transport: &http.Transport{DisableKeepAlives: true}}
+	resp, err := client.Do(req)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+	rec := httptest.NewRecorder()
+	rec.Code = resp.StatusCode
+	data := map[string]any{}
+	_ = json.NewDecoder(resp.Body).Decode(&data)
+	return rec, data
+}
+
+// freePort reserves an ephemeral port and returns it. The listener is closed
+// immediately; the tiny reuse window is acceptable for tests and each test
+// gets a different port, so nothing can collide or share pooled connections.
+func freePort(t *testing.T) int {
+	t.Helper()
+	listener, err := (&net.ListenConfig{}).Listen(context.Background(), "tcp", "127.0.0.1:0")
+	if err != nil {
+		t.Fatal(err)
+	}
+	port := listener.Addr().(*net.TCPAddr).Port
+	_ = listener.Close()
+	return port
+}
+
+func mustJSON(t *testing.T, v any) string {
+	t.Helper()
+	raw, err := json.Marshal(v)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return string(raw)
+}
+
+func TestWizardSavesConfigWithStrictPermissions(t *testing.T) {
+	sup := &fakeSup{}
+	server, base := newTestServer(t, sup)
+
+	rec, _ := postJSON(t, base, "/api/config", map[string]any{
+		"coordinator_url": "http://192.168.1.10:8080",
+		"token":           "sm_live_secret",
+		"work_dir":        "/home/emil/scimesh-worker",
+		"worker_name":     "emil-laptop",
+		"cpu_count":       8,
+		"memory_mb":       16384,
+	})
+	if rec.Code != http.StatusOK {
+		t.Fatalf("save config: got %d, want 200", rec.Code)
+	}
+	info, err := os.Stat(server.cfgPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if perm := info.Mode().Perm(); perm != 0o600 {
+		t.Errorf("config perms = %o, want 600", perm)
+	}
+	config, err := agent.LoadConfigFile(server.cfgPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if config.CoordinatorURL != "http://192.168.1.10:8080" || config.Token != "sm_live_secret" || config.WorkDir != "/home/emil/scimesh-worker" || config.WorkerName != "emil-laptop" {
+		t.Errorf("config = %+v", config)
+	}
+	if config.CPUCount != 8 || config.MemoryMB != 16384 {
+		t.Errorf("resources: cpu=%d mem=%d", config.CPUCount, config.MemoryMB)
+	}
+}
+
+func TestWizardRejectsInvalidConfig(t *testing.T) {
+	sup := &fakeSup{}
+	_, base := newTestServer(t, sup)
+
+	for _, body := range []map[string]any{
+		{"coordinator_url": "", "token": "x"},
+		{"coordinator_url": "not-a-url", "token": "x"},
+	} {
+		rec, _ := postJSON(t, base, "/api/config", body)
+		if rec.Code != http.StatusBadRequest {
+			t.Errorf("body %v: got %d, want 400", body, rec.Code)
+		}
+	}
+}
+
+func TestWizardStartStopLifecycle(t *testing.T) {
+	sup := &fakeSup{}
+	_, base := newTestServer(t, sup)
+
+	// Starting without a saved config is rejected.
+	rec, _ := postJSON(t, base, "/api/start", map[string]any{})
+	if rec.Code != http.StatusBadRequest {
+		t.Errorf("start without config: got %d, want 400", rec.Code)
+	}
+
+	postJSON(t, base, "/api/config", map[string]any{
+		"coordinator_url": "http://127.0.0.1:8080", "token": "t", "work_dir": ".",
+	})
+	rec, data := postJSON(t, base, "/api/start", map[string]any{})
+	if rec.Code != http.StatusOK || int(data["pid"].(float64)) != 4242 {
+		t.Errorf("start: got %d %v, want 200 pid 4242", rec.Code, data)
+	}
+	if !sup.started {
+		t.Error("supervisor never started the worker")
+	}
+
+	// Status reflects the running state.
+	req, _ := http.NewRequestWithContext(context.Background(), http.MethodGet, base+"/api/status", nil)
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+	var status map[string]any
+	_ = json.NewDecoder(resp.Body).Decode(&status)
+	if status["running"] != true || status["pid"] != float64(4242) {
+		t.Errorf("status = %v, want running pid 4242", status)
+	}
+
+	rec, _ = postJSON(t, base, "/api/stop", map[string]any{})
+	if rec.Code != http.StatusOK || !sup.stopped {
+		t.Errorf("stop: got %d stopped=%v, want 200/true", rec.Code, sup.stopped)
+	}
+}
+
+func TestWizardStatusPrefillsSavedConfig(t *testing.T) {
+	sup := &fakeSup{}
+	_, base := newTestServer(t, sup)
+	postJSON(t, base, "/api/config", map[string]any{
+		"coordinator_url": "http://10.0.0.5:8080", "worker_key": "smk_abc", "work_dir": "/w", "worker_name": "n1",
+	})
+	req, _ := http.NewRequestWithContext(context.Background(), http.MethodGet, base+"/api/status", nil)
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+	var status struct {
+		ConfigPresent bool   `json:"config_present"`
+		WorkerName    string `json:"worker_name"`
+		Coordinator   string `json:"coordinator"`
+		TokenSet      bool   `json:"token_set"`
+	}
+	_ = json.NewDecoder(resp.Body).Decode(&status)
+	if !status.ConfigPresent || status.WorkerName != "n1" || status.Coordinator != "http://10.0.0.5:8080" || !status.TokenSet {
+		t.Errorf("status = %+v", status)
+	}
+	// The secret must never appear in the status projection.
+	if strings.Contains(strings.ToLower(mustJSON(t, status)), "smk_abc") {
+		t.Error("status leaks the worker key")
+	}
+}
+
+func TestCheckCoordinatorReachable(t *testing.T) {
+	stub := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/health" {
+			_, _ = w.Write([]byte(`{"status":"ok"}`))
+			return
+		}
+		http.NotFound(w, r)
+	}))
+	defer stub.Close()
+
+	report := agent.CheckCoordinator(context.Background(), stub.URL, 5*time.Second)
+	if !report.Coordinator.OK {
+		t.Errorf("coordinator check = %+v, want ok", report.Coordinator)
+	}
+}
+
+func TestCheckCoordinatorUnreachable(t *testing.T) {
+	report := agent.CheckCoordinator(context.Background(), "http://127.0.0.1:1", 2*time.Second)
+	if report.Coordinator.OK {
+		t.Error("unreachable coordinator reported ok")
+	}
+}
+
+func TestConfigFileDefaultsAndEnvOverride(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "config.json")
+	file := agent.ConfigFile{
+		CoordinatorURL: "http://coord:8080",
+		Token:          "file-token",
+		WorkDir:        "/w",
+		CPUCount:       4,
+	}
+	if err := agent.SaveConfigFile(path, file); err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("COORDINATOR_URL", "http://env:9090")
+	t.Setenv("WORKER_AUTH_TOKEN", "")
+	config, err := agent.LoadConfigFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if config.CoordinatorURL != "http://env:9090" {
+		t.Errorf("env must win: %s", config.CoordinatorURL)
+	}
+	if config.Token != "file-token" {
+		t.Errorf("token = %q, want the file value", config.Token)
+	}
+	if config.CPUCount != 4 {
+		t.Errorf("cpu = %d", config.CPUCount)
+	}
+}
+
+func TestInstallRuntimeCreatesVenvAndReportsPython(t *testing.T) {
+	var installedPkg string
+	sup := &fakeSup{}
+	_, base := newTestServerWithInstall(t, sup, func(ctx context.Context, venvPython, pkg string) error {
+		installedPkg = pkg
+		// Prove the venv python path really exists by creating a marker file
+		// where the real venv python would be.
+		_ = os.MkdirAll(filepath.Dir(venvPython), 0o755)
+		_ = os.WriteFile(venvPython, []byte("#!/bin/sh\nexit 0\n"), 0o755)
+		return nil
+	})
+
+	rec, data := postJSON(t, base, "/api/runtime/install", map[string]any{"scimesh_package": "/wheels/scimesh.whl"})
+	if rec.Code != http.StatusOK || data["ok"] != true {
+		t.Fatalf("install: got %d %v, want 200 ok", rec.Code, data)
+	}
+	if installedPkg != "/wheels/scimesh.whl" {
+		t.Errorf("package = %q, want the requested wheel", installedPkg)
+	}
+	if !strings.HasSuffix(data["python"].(string), "venv/bin/python") {
+		t.Errorf("python = %v, want the venv python", data["python"])
+	}
+}
+
+func TestInstallRuntimeRequiresASource(t *testing.T) {
+	sup := &fakeSup{}
+	_, base := newTestServerWithInstall(t, sup, func(ctx context.Context, venvPython, pkg string) error {
+		t.Fatal("install must not run without a package source")
+		return nil
+	})
+	// No source anywhere (SCIMESH_PIP_PACKAGE unset, request empty): 409 with
+	// guidance. The PyPI name is another project, so no silent fallback.
+	rec, data := postJSON(t, base, "/api/runtime/install", map[string]any{})
+	if rec.Code != http.StatusConflict {
+		t.Fatalf("install without source: got %d, want 409", rec.Code)
+	}
+	if !strings.Contains(data["error"].(string), "SCIMESH_PIP_PACKAGE") {
+		t.Errorf("error = %v, want a hint about SCIMESH_PIP_PACKAGE", data["error"])
+	}
+}
+
+func TestInstallRuntimeFailureIsExplained(t *testing.T) {
+	sup := &fakeSup{}
+	_, base := newTestServerWithInstall(t, sup, func(ctx context.Context, venvPython, pkg string) error {
+		return errors.New("no matching distribution found")
+	})
+	rec, data := postJSON(t, base, "/api/runtime/install", map[string]any{"scimesh_package": "/wheels/scimesh.whl"})
+	if rec.Code != http.StatusConflict {
+		t.Fatalf("install failure: got %d, want 409", rec.Code)
+	}
+	if !strings.Contains(data["error"].(string), "SCIMESH_PIP_PACKAGE") {
+		t.Errorf("error = %v, want a hint about SCIMESH_PIP_PACKAGE", data["error"])
+	}
+}
+
+func TestInstallRuntimeDownloadsReleaseWheelWhenNoSource(t *testing.T) {
+	oldVersion := agent.Version
+	agent.Version = "1.1.0-alpha.10"
+	t.Cleanup(func() { agent.Version = oldVersion })
+
+	var downloadedURL, installedPkg string
+	sup := &fakeSup{}
+	_, base := newTestServerWithInstallAndWheel(t, sup,
+		func(ctx context.Context, venvPython, pkg string) error {
+			installedPkg = pkg
+			return nil
+		},
+		func(ctx context.Context, url, dir string) (string, error) {
+			downloadedURL = url
+			return filepath.Join(dir, "scimesh-1.1.0a10-py3-none-any.whl"), nil
+		})
+
+	rec, data := postJSON(t, base, "/api/runtime/install", map[string]any{})
+	if rec.Code != http.StatusOK || data["ok"] != true {
+		t.Fatalf("install: got %d %v, want 200 ok", rec.Code, data)
+	}
+	if !strings.Contains(downloadedURL, "releases/download/v1.1.0-alpha.10/scimesh-1.1.0a10-py3-none-any.whl") {
+		t.Errorf("download url = %q, want the release wheel of this version", downloadedURL)
+	}
+	if !strings.HasSuffix(installedPkg, "scimesh-1.1.0a10-py3-none-any.whl") {
+		t.Errorf("pip received %q, want the downloaded wheel", installedPkg)
+	}
+}
+
+func TestInstallRuntimeWheelDownloadFailureIsExplained(t *testing.T) {
+	oldVersion := agent.Version
+	agent.Version = "1.1.0-alpha.10"
+	t.Cleanup(func() { agent.Version = oldVersion })
+
+	sup := &fakeSup{}
+	_, base := newTestServerWithInstallAndWheel(t, sup,
+		func(ctx context.Context, venvPython, pkg string) error { t.Fatal("pip must not run"); return nil },
+		func(ctx context.Context, url, dir string) (string, error) {
+			return "", errors.New("HTTP 404")
+		})
+
+	rec, data := postJSON(t, base, "/api/runtime/install", map[string]any{})
+	if rec.Code != http.StatusConflict {
+		t.Fatalf("got %d, want 409", rec.Code)
+	}
+	if !strings.Contains(data["error"].(string), "SCIMESH_PIP_PACKAGE") {
+		t.Errorf("error = %v, want a SCIMESH_PIP_PACKAGE hint", data["error"])
+	}
+}
+
+func TestStartPinsTheVenvTaskRunner(t *testing.T) {
+	sup := &fakeSup{}
+	server, base := newTestServer(t, sup)
+	postJSON(t, base, "/api/config", map[string]any{
+		"coordinator_url": "http://coord:8080", "token": "t", "work_dir": ".",
+	})
+	// Simulate the runtime installer: create the venv python marker.
+	venvPython := filepath.Join(server.dir, "venv", "bin", "python")
+	_ = os.MkdirAll(filepath.Dir(venvPython), 0o755)
+	_ = os.WriteFile(venvPython, []byte("#!/bin/sh\nexit 0\n"), 0o755)
+
+	rec, _ := postJSON(t, base, "/api/start", map[string]any{})
+	if rec.Code != http.StatusOK {
+		t.Fatalf("start: got %d, want 200", rec.Code)
+	}
+	config, err := agent.LoadConfigFile(server.cfgPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(config.TaskRunner) != 3 || config.TaskRunner[0] != venvPython || config.TaskRunner[1] != "-m" || config.TaskRunner[2] != "scimesh.worker.task" {
+		t.Errorf("task runner = %v, want the venv python runner", config.TaskRunner)
+	}
+}
+
+func TestSaveConfigPinsVenvRunnerWhenPresent(t *testing.T) {
+	server, base := newTestServer(t, &fakeSup{})
+	venvPython := filepath.Join(server.dir, "venv", "bin", "python")
+	_ = os.MkdirAll(filepath.Dir(venvPython), 0o755)
+	_ = os.WriteFile(venvPython, []byte("#!/bin/sh\nexit 0\n"), 0o755)
+
+	rec, _ := postJSON(t, base, "/api/config", map[string]any{
+		"coordinator_url": "http://coord:8080", "token": "t", "work_dir": ".",
+	})
+	if rec.Code != http.StatusOK {
+		t.Fatalf("config: got %d", rec.Code)
+	}
+	config, err := agent.LoadConfigFile(server.cfgPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(config.TaskRunner) != 3 || config.TaskRunner[0] != venvPython {
+		t.Errorf("task runner = %v, want the venv python", config.TaskRunner)
+	}
+}
+
+func TestTestProbesTheVenvPythonAfterInstall(t *testing.T) {
+	sup := &fakeSup{}
+	server, base := newTestServer(t, sup)
+	// The runtime installer leaves a venv python; make it a stub that reports
+	// a fake scimesh version so the preflight goes green through the venv.
+	venvPython := filepath.Join(server.dir, "venv", "bin", "python")
+	_ = os.MkdirAll(filepath.Dir(venvPython), 0o755)
+	_ = os.WriteFile(venvPython, []byte("#!/bin/sh\nif [ \"$1\" = \"-c\" ]; then echo 9.9.9-test; exit 0; fi\nexit 0\n"), 0o755)
+
+	req, _ := http.NewRequestWithContext(context.Background(), http.MethodPost, base+"/api/test", strings.NewReader(`{"coordinator_url":"http://127.0.0.1:1"}`))
+	req.Header.Set("Content-Type", "application/json")
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+	var report agent.CheckReport
+	if err := json.NewDecoder(resp.Body).Decode(&report); err != nil {
+		t.Fatal(err)
+	}
+	if !report.Scimesh.OK || report.Scimesh.Detail != "9.9.9-test" {
+		t.Errorf("scimesh check = %+v, want the venv interpreter reporting 9.9.9-test", report.Scimesh)
+	}
+}
+
+func TestParseWorkerStats(t *testing.T) {
+	logPath := filepath.Join(t.TempDir(), "worker.log")
+	content := `time=1 level=INFO msg=registered worker_id=w1
+time=2 level=INFO msg="task claimed" task_id=t1 attempt=0
+time=3 level=INFO msg="task completed" task_id=t1 elapsed_seconds=2
+time=4 level=INFO msg="task claimed" task_id=t2 attempt=0
+time=5 level=WARN msg="task failed" task_id=t2 error_code=X retryable=true
+time=6 level=WARN msg="agent cycle failed" error="boom"
+`
+	if err := os.WriteFile(logPath, []byte(content), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	stats := parseWorkerStats(logPath)
+	if !stats.Registered || stats.Claimed != 2 || stats.Completed != 1 || stats.Failed != 1 {
+		t.Errorf("stats = %+v, want registered claimed=2 completed=1 failed=1", stats)
+	}
+}
@@ -0,0 +1,336 @@
+<!doctype html>
+<html lang="en">
+<head>
+<meta charset="utf-8">
+<meta name="viewport" content="width=device-width, initial-scale=1">
+<title>SciMesh Worker · Setup</title>
+<style>
+:root{--bg:#0b0e13;--panel:#11151d;--panel-2:#161c27;--border:#1f2634;--border-soft:#181f2b;--text:#e9ecf3;--text-2:#98a2b5;--text-3:#5d6879;--accent:#5b8cff;--accent-soft:#5b8cff1f;--accent-strong:#3f6fe0;--green:#3fce8a;--green-soft:#3fce8a1a;--amber:#e5b64f;--amber-soft:#e5b64f1a;--red:#f2647c;--red-soft:#f2647c1a;--mono:ui-monospace,"SF Mono",SFMono-Regular,Menlo,monospace;color-scheme:dark}
+*{box-sizing:border-box;margin:0;padding:0}
+body{background:radial-gradient(900px 500px at 50% -180px,#16233d66,transparent),var(--bg);color:var(--text);font:14px/1.55 Inter,-apple-system,"Segoe UI",Roboto,sans-serif;-webkit-font-smoothing:antialiased;min-height:100vh}
+button{font:inherit;color:inherit;background:none;border:0;cursor:pointer}
+input{font:inherit;color:var(--text);background:var(--panel-2);border:1px solid var(--border);border-radius:9px;padding:10px 13px;width:100%;outline:none;transition:border-color .12s,box-shadow .12s}
+input:focus{border-color:var(--accent);box-shadow:0 0 0 3px var(--accent-soft)}
+input::placeholder{color:var(--text-3)}
+code{font-family:var(--mono);font-size:.86em}
+.shell{max-width:660px;margin:0 auto;padding:44px 22px 70px}
+.brand{display:flex;align-items:center;justify-content:center;gap:11px;margin-bottom:8px}
+.brand-mark{display:grid;place-items:center;width:34px;height:34px;border-radius:10px;background:linear-gradient(135deg,#5b8cff,#7c5cff);box-shadow:0 4px 16px #5b8cff40}
+.brand-mark svg{width:18px;height:18px;stroke:#fff}
+.brand-name{font-weight:700;font-size:16px;letter-spacing:-.01em}
+.brand-name span{color:var(--text-3);font-weight:500}
+.tagline{text-align:center;color:var(--text-3);font-size:12.5px;margin-bottom:34px}
+.tagline code{color:var(--text-2)}
+.steps{display:flex;align-items:center;justify-content:center;gap:0;margin-bottom:30px}
+.step{display:flex;flex-direction:column;align-items:center;gap:7px;width:96px}
+.step-dot{display:grid;place-items:center;width:30px;height:30px;border-radius:50%;border:1.5px solid var(--border);background:var(--panel);color:var(--text-3);font-size:12.5px;font-weight:700;transition:all .2s}
+.step-label{font-size:11px;font-weight:600;color:var(--text-3);letter-spacing:.02em}
+.step.active .step-dot{border-color:var(--accent);background:var(--accent-soft);color:var(--accent);box-shadow:0 0 0 4px #5b8cff14}
+.step.active .step-label{color:var(--text)}
+.step.done .step-dot{border-color:var(--green);background:var(--green-soft);color:var(--green)}
+.step.done .step-label{color:var(--text-2)}
+.step-line{flex:1;max-width:44px;height:1.5px;background:var(--border);margin:0 6px 22px;position:relative;overflow:hidden}
+.step-line.done:after{content:"";position:absolute;inset:0;background:var(--green)}
+.card{background:var(--panel);border:1px solid var(--border-soft);border-radius:15px;padding:26px 28px;box-shadow:0 24px 60px #0000004d}
+.card h1{font-size:18px;font-weight:700;letter-spacing:-.02em;margin-bottom:4px}
+.card .sub{color:var(--text-2);font-size:13px;margin-bottom:22px}
+.field{margin-bottom:16px}
+.field label{display:block;font-size:12px;font-weight:650;letter-spacing:.04em;text-transform:uppercase;color:var(--text-3);margin-bottom:7px}
+.field .hint{margin-top:6px;font-size:12px;color:var(--text-3)}
+.field .hint code{color:var(--text-2)}
+.radio-grid{display:grid;grid-template-columns:1fr 1fr;gap:10px}
+.radio-card{border:1px solid var(--border);border-radius:11px;padding:13px 14px;cursor:pointer;transition:all .13s;background:var(--panel-2)}
+.radio-card:hover{border-color:#2a3446}
+.radio-card.sel{border-color:var(--accent);background:var(--accent-soft);box-shadow:0 0 0 3px #5b8cff14}
+.radio-card b{display:flex;align-items:center;gap:8px;font-size:13.5px}
+.radio-card b svg{width:15px;height:15px;stroke:var(--accent)}
+.radio-card p{margin-top:4px;font-size:12px;color:var(--text-2)}
+.check-row{display:flex;align-items:center;gap:12px;padding:11px 14px;border:1px solid var(--border-soft);border-radius:10px;margin-bottom:9px;background:var(--panel-2)}
+.check-ic{display:grid;place-items:center;width:24px;height:24px;border-radius:50%;flex:none}
+.check-ic svg{width:13px;height:13px;stroke-width:2.6}
+.check-ok{background:var(--green-soft)}.check-ok svg{stroke:var(--green)}
+.check-bad{background:var(--red-soft)}.check-bad svg{stroke:var(--red)}
+.check-wait{background:#ffffff10}.check-wait svg{stroke:var(--text-3)}
+.check-row b{font-size:13.5px;font-weight:600}
+.check-row span{display:block;font-size:12px;color:var(--text-3)}
+.check-row .ms{margin-left:auto;font:11.5px var(--mono);color:var(--text-3)}
+.actions{display:flex;align-items:center;justify-content:space-between;gap:12px;margin-top:24px}
+.btn{display:inline-flex;align-items:center;gap:8px;border-radius:9px;padding:10px 18px;font-weight:650;font-size:13.5px;border:1px solid transparent;transition:all .13s}
+.btn svg{width:15px;height:15px;stroke:currentColor}
+.btn-primary{background:var(--accent);color:#0a1222}
+.btn-primary:hover{background:var(--accent-strong);color:#fff}
+.btn-ghost{border-color:var(--border);color:var(--text-2);background:var(--panel-2)}
+.btn-ghost:hover{color:var(--text);border-color:#2a3446}
+.btn-danger{background:var(--red-soft);color:var(--red)}
+.btn-lg{padding:12px 24px;font-size:14.5px;border-radius:10px}
+.link{color:var(--text-3);font-size:13px}
+.link:hover{color:var(--text)}
+.status-head{display:flex;align-items:center;gap:14px;margin-bottom:22px}
+.pulse{position:relative;width:12px;height:12px;border-radius:50%;background:var(--green);flex:none}
+.pulse:after{content:"";position:absolute;inset:-5px;border-radius:50%;border:2px solid var(--green);opacity:.5;animation:ping 1.6s ease-out infinite}
+@keyframes ping{from{transform:scale(.6);opacity:.7}to{transform:scale(1.4);opacity:0}}
+.status-head h1{font-size:19px}
+.status-head .sub{font-size:12.5px;color:var(--text-3)}
+.stat-row{display:grid;grid-template-columns:repeat(4,1fr);gap:10px;margin-bottom:18px}
+.stat{background:var(--panel-2);border:1px solid var(--border-soft);border-radius:11px;padding:12px 14px}
+.stat b{display:block;font-size:20px;font-weight:700;letter-spacing:-.02em}
+.stat span{font-size:11px;font-weight:600;letter-spacing:.05em;text-transform:uppercase;color:var(--text-3)}
+.stat.bad b{color:var(--red)}
+.logbox{background:#0a0d12;border:1px solid var(--border-soft);border-radius:11px;padding:14px 16px;font:12px/1.7 var(--mono);color:#8fa3bf;max-height:210px;overflow-y:auto;white-space:pre-wrap;word-break:break-word}
+.meta-line{display:flex;gap:8px;flex-wrap:wrap;margin-bottom:16px}
+.chip{display:inline-flex;align-items:center;gap:6px;border:1px solid var(--border);border-radius:99px;padding:4px 11px;font-size:12px;color:var(--text-2);background:var(--panel-2)}
+.chip svg{width:12px;height:12px;stroke:var(--text-3)}
+.wizard-page{display:none}.wizard-page.active{display:block;animation:fade .18s ease}
+@keyframes fade{from{opacity:0;transform:translateY(5px)}to{opacity:1}}
+.error-strip{background:var(--red-soft);border:1px solid #f2647c33;border-radius:9px;padding:9px 12px;font-size:12.5px;color:#ffb3c0;margin-bottom:14px}
+.spinner{display:inline-block;width:13px;height:13px;border:2px solid var(--text-3);border-top-color:transparent;border-radius:50%;animation:spin .7s linear infinite;vertical-align:-2px;margin-right:7px}
+@keyframes spin{to{transform:rotate(360deg)}}
+.checks{padding-bottom:6px}
+</style>
+</head>
+<body>
+<div class="shell">
+  <div class="brand">
+    <div class="brand-mark"><svg viewBox="0 0 24 24" fill="none" stroke-width="2" stroke-linecap="round"><circle cx="6" cy="6" r="2.4"/><circle cx="18" cy="7" r="2.4"/><circle cx="12" cy="17" r="2.4"/><path d="M8 7.5l7.6-1M7 8.3l3.7 6.5M16.6 9.2l-3.2 5.6"/></svg></div>
+    <div class="brand-name">SciMesh <span>· Worker setup</span></div>
+  </div>
+  <p class="tagline">Local wizard served by <code>worker-agent setup</code> · <code>127.0.0.1</code></p>
+
+  <!-- ═══ WIZARD VIEW ═══ -->
+  <div id="view-wizard">
+    <div class="steps" id="steps">
+      <div class="step active" id="st1"><div class="step-dot">1</div><div class="step-label">Connect</div></div>
+      <div class="step-line" id="sl1"></div>
+      <div class="step" id="st2"><div class="step-dot">2</div><div class="step-label">Machine</div></div>
+      <div class="step-line" id="sl2"></div>
+      <div class="step" id="st3"><div class="step-dot">3</div><div class="step-label">Check</div></div>
+      <div class="step-line" id="sl3"></div>
+      <div class="step" id="st4"><div class="step-dot">4</div><div class="step-label">Run</div></div>
+    </div>
+    <div id="error-box"></div>
+
+    <!-- step 1: connect -->
+    <div class="wizard-page active" id="wp1">
+      <div class="card">
+        <h1>Connect to a coordinator</h1>
+        <p class="sub">The coordinator hands out work and collects results. Ask your cluster admin for its address.</p>
+        <div class="field"><label>Coordinator URL</label><input id="in-url" placeholder="http://192.168.1.10:8080"><p class="hint">For a served instance this is the address printed by <code>coordinator serve</code>.</p></div>
+        <div class="field"><label>Authentication</label>
+          <div class="radio-grid" id="auth-grid">
+            <div class="radio-card sel" data-mode="token"><b><svg viewBox="0 0 24 24" fill="none" stroke-width="2"><rect x="4" y="10" width="16" height="10" rx="2"/><path d="M8 10V7a4 4 0 0 1 8 0v3"/></svg>Cluster token</b><p>Serve instances: one token for every worker.</p></div>
+            <div class="radio-card" data-mode="key"><b><svg viewBox="0 0 24 24" fill="none" stroke-width="2"><circle cx="8" cy="14" r="4"/><path d="M10.8 11.2L20 2M15 4l3 3"/></svg>Worker key</b><p>Shared clusters: a key tied to your account.</p></div>
+          </div>
+        </div>
+        <div class="field" id="token-field"><label>Token</label><input id="in-token" type="password" placeholder="paste the token from coordinator token"><p class="hint">The admin can copy it from <code>coordinator token</code> on the server.</p></div>
+        <div class="field" id="key-field" style="display:none"><label>Worker key + userservice URL</label><input id="in-key" type="password" placeholder="smk_…"><input id="in-users" placeholder="http://userservice-host:8081" style="margin-top:8px"><p class="hint">Create a key in the coordinator UI: Users → worker keys.</p></div>
+        <div class="actions"><span class="link" id="l1">Step 1 of 4</span><button class="btn btn-primary" id="b1">Continue →</button></div>
+      </div>
+    </div>
+
+    <!-- step 2: machine -->
+    <div class="wizard-page" id="wp2">
+      <div class="card">
+        <h1>This machine</h1>
+        <p class="sub">Where tasks run and how the machine appears in the cluster.</p>
+        <div class="field"><label>Worker name</label><input id="in-name" placeholder="auto-detected"><p class="hint">Shown in the coordinator’s worker list.</p></div>
+        <div class="field"><label>Work directory</label><input id="in-dir" placeholder="./scimesh-agent-data"><p class="hint">Datasets and shard results live here. ~1 GB free space recommended.</p></div>
+        <div class="field"><label>Compute resources advertised</label>
+          <div class="radio-grid">
+            <div class="radio-card sel" data-cpu="auto"><b>Auto</b><p>Detect the machine’s CPU count.</p></div>
+            <div class="radio-card" data-cpu="custom"><b>Custom…</b><p>Limit what this machine advertises.</p></div>
+          </div>
+        </div>
+        <div class="field" id="cpu-field" style="display:none"><label>CPU count</label><input id="in-cpu" type="number" min="1" value="1"></div>
+        <div class="actions"><button class="btn btn-ghost" id="b2b">← Back</button><button class="btn btn-primary" id="b2">Continue →</button></div>
+      </div>
+    </div>
+
+    <!-- step 3: preflight -->
+    <div class="wizard-page" id="wp3">
+      <div class="card">
+        <h1>Preflight check</h1>
+        <p class="sub">Making sure this machine can reach the coordinator and run SciMesh workloads.</p>
+        <div class="checks" id="checks"></div>
+        <div class="actions"><button class="btn btn-ghost" id="b3b">← Back</button><button class="btn btn-primary" id="b3" disabled>Continue anyway →</button></div>
+      </div>
+    </div>
+
+    <!-- step 4: run -->
+    <div class="wizard-page" id="wp4">
+      <div class="card" style="text-align:center;padding:40px 28px">
+        <div class="brand-mark" style="margin:0 auto 18px;width:46px;height:46px;border-radius:13px"><svg viewBox="0 0 24 24" fill="none" stroke-width="2" stroke-linecap="round" style="width:22px;height:22px"><path d="M6 4l14 8-14 8V4z"/></svg></div>
+        <h1 style="font-size:20px">Ready to join the cluster</h1>
+        <p class="sub" style="max-width:380px;margin:8px auto 26px">Configuration will be saved and the worker started as a background process.</p>
+        <div class="actions" style="justify-content:space-between;margin-top:30px"><button class="btn btn-ghost" id="b4b">← Back</button><button class="btn btn-primary btn-lg" id="b4">Start worker</button></div>
+      </div>
+    </div>
+  </div>
+
+  <!-- ═══ STATUS VIEW ═══ -->
+  <div id="view-status" style="display:none">
+    <div class="card">
+      <div class="status-head">
+        <div class="pulse" id="st-pulse"></div>
+        <div><h1 id="st-title">Worker is working</h1><div class="sub" id="st-sub">—</div></div>
+        <button class="btn btn-danger" id="st-stop" style="margin-left:auto"><svg viewBox="0 0 24 24" fill="none" stroke-width="2"><rect x="6" y="6" width="12" height="12" rx="2"/></svg>Stop</button>
+      </div>
+      <div class="stat-row">
+        <div class="stat"><b id="st-claimed">—</b><span>claimed</span></div>
+        <div class="stat"><b id="st-completed">—</b><span>completed</span></div>
+        <div class="stat bad"><b id="st-failed">—</b><span>failed</span></div>
+        <div class="stat"><b id="st-registered">—</b><span>registered</span></div>
+      </div>
+      <div class="meta-line" id="st-meta"></div>
+      <div class="logbox" id="st-log"></div>
+      <div class="actions"><span class="link" id="st-cfg">—</span><button class="btn btn-ghost" id="st-reconfig">Reconfigure…</button></div>
+    </div>
+  </div>
+</div>
+
+<script>
+const $=id=>document.getElementById(id);
+let state={mode:'token',cpu:'auto',venvPython:null};
+let checksOk=false;
+
+function err(msg){$('error-box').innerHTML=msg?'<div class="error-strip">'+msg+'</div>':''}
+function goto(n){
+  ['wp1','wp2','wp3','wp4'].forEach((id,i)=>$(id).classList.toggle('active',i===n-1));
+  for(let i=1;i<=4;i++){
+    const st=$('st'+i);
+    st.classList.toggle('done',i<n);st.classList.toggle('active',i===n);
+    st.querySelector('.step-dot').textContent=i<n?'✓':i;
+    if(i<4)$('sl'+i).classList.toggle('done',i<n);
+  }
+  err('');
+}
+document.querySelectorAll('#auth-grid .radio-card').forEach(c=>c.addEventListener('click',()=>{
+  document.querySelectorAll('#auth-grid .radio-card').forEach(x=>x.classList.remove('sel'));
+  c.classList.add('sel');state.mode=c.dataset.mode;
+  $('token-field').style.display=state.mode==='token'?'':'none';
+  $('key-field').style.display=state.mode==='key'?'':'none';
+}));
+document.querySelectorAll('#wp2 .radio-card').forEach(c=>c.addEventListener('click',()=>{
+  c.parentElement.querySelectorAll('.radio-card').forEach(x=>x.classList.remove('sel'));
+  c.classList.add('sel');state.cpu=c.dataset.cpu;
+  $('cpu-field').style.display=state.cpu==='custom'?'':'none';
+}));
+
+async function postJSON(path,body){
+  const r=await fetch(path,{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify(body)});
+  const data=await r.json().catch(()=>({}));
+  return {status:r.status,data};
+}
+function draftConfig(){
+  const cfg={
+    coordinator_url:$('in-url').value.trim(),
+    token:state.mode==='token'?$('in-token').value.trim():'',
+    worker_key:state.mode==='key'?$('in-key').value.trim():'',
+    userservice_url:state.mode==='key'?$('in-users').value.trim():'',
+    work_dir:$('in-dir').value.trim(),
+    worker_name:$('in-name').value.trim(),
+    cpu_count:state.cpu==='custom'?parseInt($('in-cpu').value||'1',10):0
+  };
+  if(state.venvPython)cfg.task_runner=[state.venvPython,'-m','scimesh.worker.task'];
+  return cfg;
+}
+
+$('b1').onclick=()=>{
+  const c=draftConfig();
+  if(!c.coordinator_url){err('Enter the coordinator URL.');return}
+  if(state.mode==='token'&&!c.token){err('Enter the cluster token.');return}
+  if(state.mode==='key'&&!c.worker_key){err('Enter the worker key.');return}
+  goto(2);
+};
+$('b2b').onclick=()=>goto(1);
+$('b2').onclick=()=>{goto(3);runChecks()};
+$('b3b').onclick=()=>goto(2);
+$('b3').onclick=()=>goto(4);
+$('b4b').onclick=()=>goto(3);
+$('b4').onclick=async()=>{
+  const c=draftConfig();
+  const r=await postJSON('/api/config',c);
+  if(r.status!==200){err('Could not save the configuration: '+(r.data.error||'unknown error'));return}
+  const s=await postJSON('/api/start',{});
+  if(s.status!==200){err('Could not start the worker: '+(s.data.error||'unknown error'));return}
+  showStatus();
+};
+
+const checkRow=(name,ok,detail,ms,action)=>
+  '<div class="check-row"><div class="check-ic '+(ok===null?'check-wait':ok?'check-ok':'check-bad')+'"><svg viewBox="0 0 24 24" fill="none" stroke-linecap="round">'+(ok===null?'<path d="M12 7v5l3 3"/>':ok?'<path d="M4 12l5 5L20 6"/>':'<path d="M6 6l12 12M18 6L6 18"/>')+'</svg></div><div><b>'+name+'</b><span>'+(detail||'')+'</span></div>'+(ms?'<span class="ms">'+ms+' ms</span>':'')+(action?action:'')+'</div>';
+async function runChecks(){
+  const box=$('checks');
+  box.innerHTML=checkRow('Coordinator reachable','',null,null)+checkRow('Python 3','',null,null)+checkRow('scimesh package','',null,null);
+  const r=await postJSON('/api/test',draftConfig());
+  box.innerHTML='';
+  checksOk=true;
+  const items=[r.data.coordinator,r.data.python,r.data.scimesh];
+  for(const item of items){
+    if(item&&!item.ok)checksOk=false;
+    let action='';
+    if(item&&item.name==='scimesh'&&!item.ok){
+      action='<div style="margin-left:auto;display:flex;gap:8px;align-items:center"><input id="in-pkg" placeholder="wheel path / checkout / index URL (optional)" style="width:230px;padding:7px 10px;font-size:12px"><button class="btn btn-primary" style="padding:6px 12px;font-size:12px" id="install-scimesh">Install</button></div>';
+    }
+    box.insertAdjacentHTML('beforeend',checkRow(item?item.name:'?',item?item.ok:null,item?item.detail:'',item?item.latency_ms:null,action));
+  }
+  const btn=$('install-scimesh');
+  if(btn)btn.onclick=async()=>{
+    const src=$('in-pkg')?$('in-pkg').value.trim():'';
+    btn.disabled=true;btn.textContent='Installing…';
+    const ir=await postJSON('/api/runtime/install',{scimesh_package:src});
+    if(ir.status!==200){err('Could not install scimesh: '+(ir.data.error||'unknown error'));btn.disabled=false;btn.textContent='Install';return}
+    state.venvPython=ir.data.python;
+    runChecks();
+  };
+  $('b3').disabled=!checksOk;
+}
+
+async function showStatus(){
+  $('view-wizard').style.display='none';
+  $('view-status').style.display='block';
+  await refreshStatus();
+  setInterval(refreshStatus,2000);
+}
+async function refreshStatus(){
+  const r=await fetch('/api/status');
+  if(!r.ok)return;
+  const v=await r.json();
+  $('st-pulse').style.background=v.running?'var(--green)':'var(--text-3)';
+  $('st-pulse').style.animation=v.running?'':'none';
+  $('st-title').textContent=v.running?(v.worker_name||'Worker')+' is working':(v.worker_name||'Worker')+' is stopped';
+  $('st-sub').textContent='pid '+(v.pid||'—')+' · config '+(v.config_present?v.config_path:'not saved yet');
+  $('st-cfg').textContent='Configuration: '+(v.config_present?v.config_path:'—');
+  const stats=v.stats||{};
+  $('st-claimed').textContent=stats.claimed!=null?stats.claimed:'—';
+  $('st-completed').textContent=stats.completed!=null?stats.completed:'—';
+  $('st-failed').textContent=stats.failed!=null?stats.failed:'—';
+  $('st-registered').textContent=stats.registered?'yes':'no';
+  const meta=$('st-meta');
+  meta.innerHTML='';
+  if(v.coordinator){const c=document.createElement('span');c.className='chip';c.innerHTML='<svg viewBox="0 0 24 24" fill="none" stroke-width="2"><circle cx="12" cy="12" r="9"/></svg>'+v.coordinator;meta.append(c)}
+  if(v.work_dir){const d=document.createElement('span');d.className='chip';d.innerHTML='<svg viewBox="0 0 24 24" fill="none" stroke-width="2"><path d="M4 6h16M4 12h16M4 18h16"/></svg>'+v.work_dir;meta.append(d)}
+  if(!v.token_set){const t=document.createElement('span');t.className='chip';t.style.color='var(--amber)';t.textContent='no credential set';meta.append(t)}
+  const logs=await fetch('/api/logs?tail=200');
+  const lv=await logs.json();
+  $('st-log').textContent=lv.log||'(no log yet — the worker writes here once started)';
+}
+$('st-stop').onclick=async()=>{await postJSON('/api/stop',{});refreshStatus()};
+$('st-reconfig').onclick=()=>{
+  $('view-status').style.display='none';
+  $('view-wizard').style.display='block';
+  goto(1);
+};
+
+// Prefill from a saved configuration, then decide which view to show.
+(async()=>{
+  const r=await fetch('/api/status');
+  const v=await r.json();
+  if(v.config_present){
+    $('in-url').value=v.coordinator||'';
+    $('in-dir').value=v.work_dir||'';
+    $('in-name').value=v.worker_name||'';
+  }
+  if(v.running)showStatus();
+})();
+</script>
+</body>
+</html>
@@ -0,0 +1,86 @@
+package agent
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"os"
+	"os/exec"
+	"path/filepath"
+)
+
+// TaskRunner spawns the Python task entry and returns the sealed partial
+// artifact manifest it produced.
+type TaskRunner struct {
+	command []string
+}
+
+func NewTaskRunner(command []string) *TaskRunner {
+	return &TaskRunner{command: command}
+}
+
+// Run executes one task: the task payload is written as JSON into the attempt
+// directory, the Python entry computes and seals the partial, and the written
+// manifest is parsed back. stderr is captured for failure reporting.
+func (r *TaskRunner) Run(task *Task, taskDir string, manifestPath string, extraEnv []string) (*TaskRunnerManifest, error) {
+	if err := os.MkdirAll(taskDir, 0o750); err != nil {
+		return nil, err
+	}
+	payload := map[string]any{
+		"task_id":          task.TaskID,
+		"attempt":          task.Attempt,
+		"lease_expires_at": task.leaseExpiresRaw,
+		"workload":         task.Workload,
+		"input": map[string]any{
+			"uri":    task.Input.URI,
+			"sha256": task.Input.SHA256,
+		},
+		"parameters": task.Parameters,
+	}
+	payloadBytes, err := json.Marshal(payload)
+	if err != nil {
+		return nil, err
+	}
+	taskJSONPath := filepath.Join(taskDir, "task.json")
+	if err := os.WriteFile(taskJSONPath, payloadBytes, 0o600); err != nil {
+		return nil, err
+	}
+	args := append([]string{}, r.command[1:]...)
+	args = append(args,
+		"--task-json", taskJSONPath,
+		"--task-dir", taskDir,
+		"--output", manifestPath,
+	)
+	// #nosec G204 -- the command comes from the operator-configured TASK_RUNNER.
+	command := exec.CommandContext(context.Background(), r.command[0], args...)
+	command.Dir = taskDir
+	command.Env = append(os.Environ(), extraEnv...)
+	var stderr bytes.Buffer
+	command.Stderr = &stderr
+	if err := command.Run(); err != nil {
+		var exitErr *exec.ExitError
+		if errors.As(err, &exitErr) {
+			return nil, runnerExitError(exitErr.ExitCode(), stderr.String())
+		}
+		return nil, fmt.Errorf("task runner could not be started: %w", err)
+	}
+	// #nosec G304 -- the manifest path is inside the worker's own task directory.
+	raw, err := os.ReadFile(manifestPath)
+	if err != nil {
+		return nil, fmt.Errorf("task runner produced no result manifest")
+	}
+	var manifest TaskRunnerManifest
+	if err := json.Unmarshal(raw, &manifest); err != nil {
+		return nil, fmt.Errorf("task runner produced an invalid result manifest")
+	}
+	if manifest.ArtifactPath == "" || manifest.ContentType == "" {
+		return nil, fmt.Errorf("task runner produced an incomplete result manifest")
+	}
+	info, err := os.Stat(manifest.ArtifactPath)
+	if err != nil || !info.Mode().IsRegular() {
+		return nil, fmt.Errorf("task runner produced no artifact file")
+	}
+	return &manifest, nil
+}
@@ -18,4 +18,5 @@ var (
 	ErrResultConflict    = errors.New("different result already recorded")
 	ErrInvalidInput      = errors.New("invalid input")
 	ErrTaskNotLeased     = errors.New("task is not currently leased")
+	ErrWorkloadDisabled  = errors.New("workload is disabled")
 )
@@ -1,6 +1,7 @@
 package domain

 import (
+	"strings"
 	"time"

 	"github.com/google/uuid"
@@ -52,6 +53,10 @@ func NewWorker(name string, capabilities []string, now time.Time) (*Worker, erro
 	if len(capabilities) == 0 {
 		return nil, ErrInvalidInput
 	}
+	name = strings.TrimSpace(name)
+	if name == "" {
+		return nil, ErrInvalidInput
+	}
 	return &Worker{
 		ID:              uuid.New(),
 		Name:            name,
@@ -29,3 +29,11 @@ func TestNewWorkerRejectsNoCapabilities(t *testing.T) {
 		t.Errorf("empty slice: err = %v, want ErrInvalidInput", err)
 	}
 }
+
+func TestNewWorkerRejectsBlankName(t *testing.T) {
+	for _, name := range []string{"", "   ", "\t\n"} {
+		if _, err := NewWorker(name, []string{"similarity-search"}, testNow); !errors.Is(err, ErrInvalidInput) {
+			t.Errorf("name %q: got %v, want ErrInvalidInput", name, err)
+		}
+	}
+}
@@ -8,6 +8,7 @@ import (
 	"io/fs"
 	"math"
 	"os"
+	"path/filepath"
 	"strconv"
 	"time"

@@ -51,6 +52,9 @@ type Config struct {
 	LogFile string
 	// Directory where artifact bytes are stored.
 	StorageDir string
+	// Directory of the built MkDocs site (site/) served at /ui/docs/. Empty
+	// disables the docs route; the UI shows a hint page instead.
+	DocsDir string
 	// Upper bound on an uploaded dataset or artifact body, in bytes.
 	MaxUploadBytes int64

@@ -75,6 +79,16 @@ type Config struct {
 	ReaperInterval time.Duration
 	// A worker silent for longer than this is marked offline by the reaper.
 	WorkerOfflineAfter time.Duration
+	// Whether the binary applies its embedded schema migrations on startup.
+	// On by default so a downloaded binary provisions its own database; set
+	// AUTO_MIGRATE=false when an operator manages migrations out of band.
+	AutoMigrate bool
+	// DatabaseEngine selects the storage backend: "sqlite" (embedded, the
+	// single-binary default) or "postgres" (cluster deployments). The
+	// postgres engine requires DATABASE_URL.
+	DatabaseEngine string
+	// DBPath is the sqlite database file (engine=sqlite only).
+	DBPath string
 }

 // Load reads the environment and fails fast on anything required-but-missing
@@ -108,6 +122,7 @@ func LoadConfig() (Config, error) {
 		LogLevel:             getEnv("LOG_LEVEL", "info"),
 		LogFile:              os.Getenv("LOG_FILE"),
 		StorageDir:           getEnv("COORDINATOR_STORAGE_DIR", "./data"),
+		DocsDir:              os.Getenv("SCIMESH_DOCS_DIR"),
 		MaxUploadBytes:       1 << 30, // 1 GiB
 		DBMaxConns:           10,
 		DBConnectTimeout:     30 * time.Second,
@@ -120,8 +135,16 @@ func LoadConfig() (Config, error) {
 		WorkerOfflineAfter:   1 * time.Minute,
 	}

-	if cfg.DatabaseURL == "" {
-		return Config{}, fmt.Errorf("DATABASE_URL is required")
+	cfg.DatabaseEngine = getEnv("SCIMESH_DB", "sqlite")
+	switch cfg.DatabaseEngine {
+	case "sqlite", "postgres":
+	default:
+		return Config{}, fmt.Errorf("SCIMESH_DB must be sqlite or postgres")
+	}
+	cfg.DBPath = getEnv("SCIMESH_DB_PATH", filepath.Join(cfg.StorageDir, "scimesh.db"))
+
+	if cfg.DatabaseEngine == "postgres" && cfg.DatabaseURL == "" {
+		return Config{}, fmt.Errorf("DATABASE_URL is required for the postgres engine")
 	}
 	if cfg.UIToken != "" && cfg.Token != "" && cfg.UIToken == cfg.Token {
 		return Config{}, fmt.Errorf("UI_AUTH_TOKEN must differ from the worker auth token")
@@ -169,7 +192,14 @@ func LoadConfig() (Config, error) {
 	if cfg.DefaultMaxAttempts < 1 {
 		return Config{}, fmt.Errorf("DEFAULT_MAX_ATTEMPTS must be positive")
 	}
-
+	cfg.AutoMigrate = true
+	if raw := os.Getenv("AUTO_MIGRATE"); raw != "" {
+		parsed, err := strconv.ParseBool(raw)
+		if err != nil {
+			return Config{}, fmt.Errorf("AUTO_MIGRATE must be true or false")
+		}
+		cfg.AutoMigrate = parsed
+	}
 	return cfg, nil
 }

@@ -314,6 +314,17 @@ func (r *WorkerRepo) MarkStaleOffline(ctx context.Context, cutoff time.Time) (in
 	return n, nil
 }

+func (r *WorkerRepo) SetTrust(ctx context.Context, id uuid.UUID, trust domain.WorkerTrust) error {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	w, ok := r.workers[id]
+	if !ok {
+		return domain.ErrWorkerNotFound
+	}
+	w.TrustLevel = trust
+	return nil
+}
+
 // --- ArtifactRepo --------------------------------------------------------

 type ArtifactRepo struct {
@@ -450,3 +461,32 @@ func (r *TaskResultRepo) CountAgreeing(_ context.Context, taskID uuid.UUID, sha2
 	}
 	return n, nil
 }
+
+func (r *JobRepo) ListCompletedBefore(ctx context.Context, cutoff time.Time) ([]domain.Job, error) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	var out []domain.Job
+	for _, j := range r.jobs {
+		if j.CompletedAt != nil && j.CompletedAt.Before(cutoff) {
+			out = append(out, *j)
+		}
+	}
+	return out, nil
+}
+
+func (r *JobRepo) Delete(ctx context.Context, id uuid.UUID) error {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	delete(r.jobs, id)
+	return nil
+}
+
+func (r *WorkerRepo) Delete(ctx context.Context, id uuid.UUID) error {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	if _, ok := r.workers[id]; !ok {
+		return domain.ErrWorkerNotFound
+	}
+	delete(r.workers, id)
+	return nil
+}
@@ -0,0 +1,57 @@
+// Package reducer contains deterministic, coordinator-side result reductions.
+package reducer
+
+import (
+	"bytes"
+	"encoding/csv"
+	"errors"
+	"fmt"
+	"io"
+)
+
+// ReduceOrderedConcat concatenates worker partial tables in shard order into a
+// single table with one header. Every partial must carry the same header as the
+// first partial and rows of the same width; anything else fails the job closed.
+func ReduceOrderedConcat(partials []io.Reader) ([]byte, error) {
+	var out bytes.Buffer
+	writer := csv.NewWriter(&out)
+	var firstHeader []string
+	for _, partial := range partials {
+		reader := csv.NewReader(partial)
+		header, err := reader.Read()
+		if err != nil {
+			return nil, fmt.Errorf("read partial header: %w", err)
+		}
+		if len(header) == 0 {
+			return nil, fmt.Errorf("partial result has an empty header")
+		}
+		if firstHeader == nil {
+			firstHeader = header
+			if err := writer.Write(header); err != nil {
+				return nil, err
+			}
+		} else if !equalStrings(header, firstHeader) {
+			return nil, fmt.Errorf("partial result has an inconsistent header")
+		}
+		for {
+			row, err := reader.Read()
+			if errors.Is(err, io.EOF) {
+				break
+			}
+			if err != nil {
+				return nil, fmt.Errorf("read partial row: %w", err)
+			}
+			if len(row) != len(header) {
+				return nil, fmt.Errorf("partial result has a row with an inconsistent width")
+			}
+			if err := writer.Write(row); err != nil {
+				return nil, err
+			}
+		}
+	}
+	writer.Flush()
+	if err := writer.Error(); err != nil {
+		return nil, err
+	}
+	return out.Bytes(), nil
+}
@@ -0,0 +1,64 @@
+package reducer
+
+import (
+	"io"
+	"strings"
+	"testing"
+)
+
+func TestReduceOrderedConcatJoinsPartialsInOrderWithOneHeader(t *testing.T) {
+	first := strings.NewReader("chembl_id,canonical_smiles\nA,CC\nB,CCC\n")
+	second := strings.NewReader("chembl_id,canonical_smiles\nC,CCCC\n")
+
+	output, err := ReduceOrderedConcat([]io.Reader{first, second})
+	if err != nil {
+		t.Fatal(err)
+	}
+	want := "chembl_id,canonical_smiles\nA,CC\nB,CCC\nC,CCCC\n"
+	if string(output) != want {
+		t.Fatalf("output = %q, want %q", output, want)
+	}
+}
+
+func TestReduceOrderedConcatIsDeterministicAcrossInputOrder(t *testing.T) {
+	left := strings.NewReader("id,rows\nA,1\nB,2\n")
+	right := strings.NewReader("id,rows\nC,3\n")
+
+	first, err := ReduceOrderedConcat([]io.Reader{left, right})
+	if err != nil {
+		t.Fatal(err)
+	}
+	left, right = strings.NewReader("id,rows\nA,1\nB,2\n"), strings.NewReader("id,rows\nC,3\n")
+	second, err := ReduceOrderedConcat([]io.Reader{left, right})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if string(first) != string(second) {
+		t.Fatalf("concat is not deterministic: %q != %q", first, second)
+	}
+}
+
+func TestReduceOrderedConcatRejectsInconsistentHeaders(t *testing.T) {
+	first := strings.NewReader("a,b\n1,2\n")
+	second := strings.NewReader("a,c\n1,2\n")
+	if _, err := ReduceOrderedConcat([]io.Reader{first, second}); err == nil {
+		t.Fatal("inconsistent headers must fail")
+	}
+}
+
+func TestReduceOrderedConcatRejectsRaggedRows(t *testing.T) {
+	partial := strings.NewReader("a,b\n1,2,3\n")
+	if _, err := ReduceOrderedConcat([]io.Reader{partial}); err == nil {
+		t.Fatal("ragged rows must fail")
+	}
+}
+
+func TestReduceOrderedConcatEmptyPartials(t *testing.T) {
+	output, err := ReduceOrderedConcat(nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(output) != 0 {
+		t.Fatalf("empty input must produce empty output, got %q", output)
+	}
+}
@@ -0,0 +1,239 @@
+// Package setup implements the `coordinator setup` wizard: database reachability
+// and creation, embedded schema migration, secret generation, and .env writing.
+// The wizard never logs or echoes secrets.
+package setup
+
+import (
+	"bufio"
+	"context"
+	"crypto/rand"
+	"encoding/hex"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+	"github.com/jackc/pgx/v5/pgconn"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/storage/postgres"
+)
+
+// Options configures one wizard run.
+type Options struct {
+	// DatabaseURL is the target coordinator database (pgx/libpq URL).
+	DatabaseURL string
+	// AdminDatabaseURL, when set, is used to create a missing target database.
+	// Defaults to the target URL with the database name replaced by "postgres".
+	AdminDatabaseURL string
+	// EnvFile is where the generated settings are written (default ".env").
+	EnvFile string
+	// Force overwrites an existing EnvFile.
+	Force bool
+	// Yes disables interactive prompts; missing values fail instead.
+	Yes bool
+	// ConnectTimeout bounds the reachability check.
+	ConnectTimeout time.Duration
+	// Out receives progress and summary output; In feeds interactive answers.
+	Out io.Writer
+	In  io.Reader
+}
+
+// Run executes the wizard and returns a summary of what was done.
+func Run(ctx context.Context, options Options) (string, error) {
+	if options.DatabaseURL == "" {
+		return "", fmt.Errorf("DATABASE_URL is required (or pass --db)")
+	}
+	if options.EnvFile == "" {
+		options.EnvFile = ".env"
+	}
+	if options.ConnectTimeout <= 0 {
+		options.ConnectTimeout = 5 * time.Second
+	}
+	if options.Out == nil {
+		options.Out = os.Stdout
+	}
+
+	report := func(format string, args ...any) {
+		_, _ = fmt.Fprintf(options.Out, format+"\n", args...)
+	}
+
+	report("SciMesh coordinator setup")
+	report("")
+
+	// 1. Reachability, with optional database creation.
+	target, err := pgx.ParseConfig(options.DatabaseURL)
+	if err != nil {
+		return "", fmt.Errorf("DATABASE_URL is not a valid postgres URL: %w", err)
+	}
+	if err := probeDatabase(ctx, target, options.ConnectTimeout); err != nil {
+		var pgErr *pgconn.PgError
+		if !errors.As(err, &pgErr) || pgErr.Code != "3D000" {
+			return "", fmt.Errorf("cannot reach the coordinator database: %w", err)
+		}
+		report("database %q does not exist yet", target.Database)
+		admin, err := resolveAdminConfig(options, target)
+		if err != nil {
+			return "", err
+		}
+		if err := createDatabase(ctx, admin, target.Database, options.ConnectTimeout); err != nil {
+			return "", fmt.Errorf("cannot create database %q: %w", target.Database, err)
+		}
+		report("created database %q", target.Database)
+	}
+	report("database %q is reachable", target.Database)
+
+	// 2. Apply the embedded schema migrations (idempotent).
+	if err := postgres.Migrate(ctx, options.DatabaseURL, nil); err != nil {
+		return "", fmt.Errorf("apply schema migrations: %w", err)
+	}
+	report("schema migrations applied")
+
+	// 3. JWT secret: reuse the environment value when strong, else generate.
+	secret := os.Getenv("JWT_SECRET")
+	if secret != "" && len(secret) < 32 {
+		return "", fmt.Errorf("JWT_SECRET must be at least 32 bytes")
+	}
+	if secret == "" {
+		generated, err := generateSecret()
+		if err != nil {
+			return "", fmt.Errorf("generate JWT_SECRET: %w", err)
+		}
+		secret = generated
+		report("generated a fresh JWT_SECRET")
+	}
+
+	// 4. Write the .env file.
+	storageDir := os.Getenv("COORDINATOR_STORAGE_DIR")
+	if storageDir == "" {
+		storageDir = "./data"
+	}
+	if err := writeEnvFile(options, secret, storageDir); err != nil {
+		return "", err
+	}
+
+	// 5. Summary.
+	var summary strings.Builder
+	fmt.Fprintf(&summary, "Setup complete.\n\n")
+	fmt.Fprintf(&summary, "Ready:\n")
+	fmt.Fprintf(&summary, "  - database %s is reachable and migrated\n", target.Database)
+	fmt.Fprintf(&summary, "  - settings written to %s (chmod 0600)\n", options.EnvFile)
+	fmt.Fprintf(&summary, "\nStart the coordinator:\n")
+	fmt.Fprintf(&summary, "  ENV_FILE=%s ./coordinator\n", options.EnvFile)
+	fmt.Fprintf(&summary, "\nOptional — userservice for UI logins (must share JWT_SECRET):\n")
+	fmt.Fprintf(&summary, "  cd users && JWT_SECRET=%q docker compose up -d\n", secret)
+	fmt.Fprintf(&summary, "  then set USERSERVICE_URL=http://localhost:8081 and BOOTSTRAP_ADMIN_EMAIL/PASSWORD\n")
+	fmt.Fprintf(&summary, "\nThe wizard cannot run PostgreSQL or the userservice for you; the\n")
+	fmt.Fprintf(&summary, "commands above are the supported way to start them.\n")
+	return summary.String(), nil
+}
+
+// probeDatabase verifies the target database accepts connections.
+func probeDatabase(ctx context.Context, config *pgx.ConnConfig, timeout time.Duration) error {
+	ctx, cancel := context.WithTimeout(ctx, timeout)
+	defer cancel()
+	conn, err := pgx.ConnectConfig(ctx, config)
+	if err != nil {
+		return err
+	}
+	return conn.Close(ctx)
+}
+
+// resolveAdminConfig picks the maintenance connection used to create
+// databases. pgx's ConnConfig.ConnString() caches the original URL, so the
+// config itself (not a re-rendered string) is what the caller connects with.
+func resolveAdminConfig(options Options, target *pgx.ConnConfig) (*pgx.ConnConfig, error) {
+	if options.AdminDatabaseURL != "" {
+		config, err := pgx.ParseConfig(options.AdminDatabaseURL)
+		if err != nil {
+			return nil, fmt.Errorf("--admin-db is not a valid postgres URL: %w", err)
+		}
+		return config, nil
+	}
+	admin := *target
+	admin.Database = "postgres"
+	return &admin, nil
+}
+
+// createDatabase creates the named database through the maintenance connection.
+func createDatabase(ctx context.Context, admin *pgx.ConnConfig, name string, timeout time.Duration) error {
+	ctx, cancel := context.WithTimeout(ctx, timeout)
+	defer cancel()
+	conn, err := pgx.ConnectConfig(ctx, admin)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = conn.Close(ctx) }()
+	quoted := `"` + strings.ReplaceAll(name, `"`, `""`) + `"`
+	if _, err := conn.Exec(ctx, "CREATE DATABASE "+quoted); err != nil {
+		return err
+	}
+	return nil
+}
+
+// generateSecret returns 32 random bytes as lowercase hex.
+func generateSecret() (string, error) {
+	buffer := make([]byte, 32)
+	if _, err := rand.Read(buffer); err != nil {
+		return "", err
+	}
+	return hex.EncodeToString(buffer), nil
+}
+
+// writeEnvFile writes the settings, refusing to clobber without --force.
+func writeEnvFile(options Options, secret, storageDir string) error {
+	path := filepath.Clean(options.EnvFile)
+	if _, err := os.Stat(path); err == nil && !options.Force {
+		return fmt.Errorf("%s already exists (use --force to overwrite)", path)
+	}
+	content := strings.Join([]string{
+		"DATABASE_URL=" + options.DatabaseURL,
+		"JWT_SECRET=" + secret,
+		"COORDINATOR_STORAGE_DIR=" + storageDir,
+		"", // trailing newline
+	}, "\n")
+	// #nosec G703 -- the env file path is operator-supplied (--env-file / ENV_FILE).
+	if err := os.WriteFile(path, []byte(content), 0o600); err != nil {
+		return fmt.Errorf("write %s: %w", path, err)
+	}
+	if err := os.Chmod(path, 0o600); err != nil {
+		return fmt.Errorf("chmod %s: %w", path, err)
+	}
+	return nil
+}
+
+// SanitizeDatabaseURL hides the password for logging.
+func SanitizeDatabaseURL(raw string) string {
+	at := strings.LastIndex(raw, "@")
+	if at < 0 {
+		return raw
+	}
+	start := 0
+	if strings.HasPrefix(raw, "postgres://") || strings.HasPrefix(raw, "postgresql://") {
+		start = len("postgres://")
+	}
+	colon := strings.Index(raw[start:at], ":")
+	if colon < 0 {
+		return raw
+	}
+	colon += start
+	return raw[:colon] + ":***@" + raw[at+1:]
+}
+
+// prompt asks a question and returns the trimmed answer ("" on EOF).
+func prompt(options Options, question, fallback string) string {
+	_, _ = fmt.Fprintf(options.Out, "%s [%s]: ", question, fallback)
+	reader := bufio.NewReader(options.In)
+	line, err := reader.ReadString('\n')
+	if err != nil && !errors.Is(err, io.EOF) {
+		return fallback
+	}
+	answer := strings.TrimSpace(line)
+	if answer == "" {
+		return fallback
+	}
+	return answer
+}
@@ -0,0 +1,91 @@
+//go:build integration
+
+package setup
+
+import (
+	"context"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+)
+
+func TestRunProvisionsDatabaseSchemaAndEnvFile(t *testing.T) {
+	ctx := context.Background()
+	base := os.Getenv("TEST_DATABASE_URL")
+	if base == "" {
+		t.Skip("TEST_DATABASE_URL is not set")
+	}
+	// The wizard must create a *missing* database through the admin URL.
+	slash := strings.LastIndex(base, "/")
+	target := base[:slash+1] + "scimesh_setup_test"
+	admin := base[:slash+1] + "postgres"
+
+	cleanup := func() {
+		conn, err := pgx.Connect(ctx, admin)
+		if err != nil {
+			return
+		}
+		defer func() { _ = conn.Close(ctx) }()
+		_, _ = conn.Exec(ctx, `DROP DATABASE IF EXISTS "scimesh_setup_test"`)
+	}
+	cleanup()
+	t.Cleanup(cleanup)
+
+	envPath := filepath.Join(t.TempDir(), ".env")
+	var output strings.Builder
+	options := Options{
+		DatabaseURL:      target,
+		AdminDatabaseURL: admin,
+		EnvFile:          envPath,
+		Force:            true,
+		Yes:              true,
+		ConnectTimeout:   10 * time.Second,
+		Out:              &output,
+		In:               strings.NewReader(""),
+	}
+	summary, err := Run(ctx, options)
+	if err != nil {
+		t.Fatalf("setup run: %v", err)
+	}
+	for _, expected := range []string{"created database", "schema migrations applied"} {
+		if !strings.Contains(output.String(), expected) {
+			t.Errorf("progress output is missing %q:\n%s", expected, output.String())
+		}
+	}
+	for _, expected := range []string{"Setup complete", "Start the coordinator"} {
+		if !strings.Contains(summary, expected) {
+			t.Errorf("summary is missing %q:\n%s", expected, summary)
+		}
+	}
+
+	// The database now exists, is migrated, and the env file is written.
+	conn, err := pgx.Connect(ctx, target)
+	if err != nil {
+		t.Fatalf("connect to provisioned database: %v", err)
+	}
+	defer func() { _ = conn.Close(ctx) }()
+	var watermark int64
+	if err := conn.QueryRow(ctx, "SELECT COALESCE(MAX(version), 0) FROM schema_migrations").Scan(&watermark); err != nil {
+		t.Fatalf("read schema_migrations: %v", err)
+	}
+	if watermark < 13 {
+		t.Errorf("schema watermark = %d, want >= 13", watermark)
+	}
+	envContent, err := os.ReadFile(envPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	env := string(envContent)
+	if !strings.Contains(env, "DATABASE_URL="+target) || !strings.Contains(env, "JWT_SECRET=") {
+		t.Errorf("env file is incomplete:\n%s", env)
+	}
+
+	// A second run is idempotent: no create-database error, same outcome.
+	if _, err := Run(ctx, options); err != nil {
+		t.Fatalf("second setup run: %v", err)
+	}
+}
@@ -0,0 +1,96 @@
+package setup
+
+import (
+	"bytes"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestGenerateSecretIsRandomAndStrong(t *testing.T) {
+	first, err := generateSecret()
+	if err != nil {
+		t.Fatal(err)
+	}
+	second, err := generateSecret()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(first) != 64 || len(second) != 64 {
+		t.Fatalf("secrets must be 32 random bytes as hex, got %d and %d", len(first), len(second))
+	}
+	if first == second {
+		t.Fatal("two generated secrets must differ")
+	}
+}
+
+func TestWriteEnvFileContentsAndPermissions(t *testing.T) {
+	path := filepath.Join(t.TempDir(), ".env")
+	if err := writeEnvFile(Options{
+		EnvFile:     path,
+		DatabaseURL: "postgres://scimesh@localhost/scimesh",
+	}, "s3cr3t", "./data"); err != nil {
+		t.Fatal(err)
+	}
+	content, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	want := "DATABASE_URL=postgres://scimesh@localhost/scimesh\nJWT_SECRET=s3cr3t\nCOORDINATOR_STORAGE_DIR=./data\n"
+	if got := string(content); got != want {
+		t.Fatalf("env file = %q, want %q", got, want)
+	}
+	info, err := os.Stat(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if info.Mode().Perm() != 0o600 {
+		t.Errorf("env file mode = %o, want 0600", info.Mode().Perm())
+	}
+}
+
+func TestWriteEnvFileRefusesWithoutForce(t *testing.T) {
+	path := filepath.Join(t.TempDir(), ".env")
+	if err := writeEnvFile(Options{EnvFile: path, DatabaseURL: "postgres://x@localhost/a"}, "a", "./data"); err != nil {
+		t.Fatal(err)
+	}
+	if err := writeEnvFile(Options{EnvFile: path, DatabaseURL: "postgres://x@localhost/a"}, "b", "./data"); err == nil {
+		t.Fatal("second write without --force must fail")
+	}
+	if err := writeEnvFile(Options{EnvFile: path, Force: true, DatabaseURL: "postgres://x@localhost/a"}, "b", "./data"); err != nil {
+		t.Fatalf("write with --force: %v", err)
+	}
+}
+
+func TestPromptReadsAnswer(t *testing.T) {
+	var out bytes.Buffer
+	answer := prompt(Options{Out: &out, In: strings.NewReader("postgres://custom\n")}, "Database URL", "default")
+	if answer != "postgres://custom" {
+		t.Fatalf("answer = %q, want the typed value", answer)
+	}
+	if !strings.Contains(out.String(), "Database URL [default]:") {
+		t.Fatalf("prompt output = %q", out.String())
+	}
+	fallback := prompt(Options{Out: &out, In: strings.NewReader("\n")}, "Question", "fb")
+	if fallback != "fb" {
+		t.Fatalf("empty answer must fall back, got %q", fallback)
+	}
+}
+
+func TestSanitizeDatabaseURL(t *testing.T) {
+	cases := map[string]string{
+		"postgres://scimesh:hunter2@localhost:5432/scimesh?sslmode=disable": "postgres://scimesh:***@localhost:5432/scimesh?sslmode=disable",
+		"postgresql://scimesh@localhost/scimesh":                            "postgresql://scimesh@localhost/scimesh",
+		"not-a-url":                                                         "not-a-url",
+	}
+	for raw, want := range cases {
+		got := SanitizeDatabaseURL(raw)
+		if got != want {
+			t.Errorf("sanitize(%q) = %q, want %q", raw, got, want)
+		}
+		if strings.Contains(got, "hunter2") {
+			t.Errorf("sanitize(%q) leaked the password", raw)
+		}
+	}
+}
@@ -0,0 +1,278 @@
+//go:build integration
+
+package postgres
+
+import (
+	"context"
+	"errors"
+	"os"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+// ensureMigrated applies the embedded schema first: the admin tests run
+// before the dedicated migration test (file order) and need real tables.
+func ensureMigrated(t *testing.T) {
+	t.Helper()
+	url := os.Getenv("TEST_DATABASE_URL")
+	if url == "" {
+		t.Skip("TEST_DATABASE_URL is not set")
+	}
+	if err := Migrate(context.Background(), url, nil); err != nil {
+		t.Fatalf("migrate: %v", err)
+	}
+}
+
+func TestAdminWorkerSetTrust(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	repo := NewWorkerRepo(pool)
+
+	w, err := domain.NewWorker("trust-lab", []string{"similarity-search"}, time.Now().UTC())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.Insert(ctx, w); err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM workers WHERE id = $1`, w.ID) })
+
+	if err := repo.SetTrust(ctx, w.ID, domain.WorkerUntrusted); err != nil {
+		t.Fatal(err)
+	}
+	got, err := repo.Get(ctx, w.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.TrustLevel != domain.WorkerUntrusted {
+		t.Errorf("trust = %q, want untrusted", got.TrustLevel)
+	}
+	if err := repo.SetTrust(ctx, uuid.New(), domain.WorkerTrusted); !errors.Is(err, domain.ErrWorkerNotFound) {
+		t.Errorf("unknown worker: got %v, want ErrWorkerNotFound", err)
+	}
+}
+
+func TestAdminListJobsPaginatedAndCounts(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	jobRepo := NewJobRepo(pool)
+	adminRepo := NewAdminReadRepo(pool)
+
+	jobs := make([]*domain.Job, 4)
+	for i := range jobs {
+		job, _ := seedJob(t, pool, 1)
+		jobs[i] = job
+	}
+	for _, j := range jobs {
+		t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM jobs WHERE id = $1`, j.ID) })
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[0].ID, domain.JobCompleted, nil); err != nil {
+		t.Fatal(err)
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[1].ID, domain.JobCompleted, nil); err != nil {
+		t.Fatal(err)
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[2].ID, domain.JobRunning, nil); err != nil {
+		t.Fatal(err)
+	}
+
+	all, total, err := adminRepo.ListJobsPaginated(ctx, "", 100, 0)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 4 || len(all) != 4 {
+		t.Errorf("all: total=%d len=%d, want 4/4", total, len(all))
+	}
+	completed, total, err := adminRepo.ListJobsPaginated(ctx, "completed", 100, 0)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 2 || len(completed) != 2 {
+		t.Errorf("completed: total=%d len=%d, want 2/2", total, len(completed))
+	}
+	page, total, err := adminRepo.ListJobsPaginated(ctx, "", 2, 2)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 4 || len(page) != 2 {
+		t.Errorf("page: total=%d len=%d, want 4/2", total, len(page))
+	}
+
+	counts, err := adminRepo.CountJobsByStatus(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if counts["completed"] != 2 || counts["running"] != 1 || counts["pending"] != 1 {
+		t.Errorf("counts = %v", counts)
+	}
+}
+
+func TestAdminTaskCountsByJobs(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	job, tasks := seedJob(t, pool, 3)
+	t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM jobs WHERE id = $1`, job.ID) })
+
+	resultArtifact := seedArtifact(t, pool, job.ID, &tasks[0].ID, domain.ArtifactPartialResult)
+	if _, err := pool.Exec(ctx, `UPDATE tasks SET status = 'completed', result_artifact_id = $1 WHERE id = $2`,
+		resultArtifact.ID, tasks[0].ID); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := pool.Exec(ctx, `UPDATE tasks SET status = 'failed' WHERE id = $1`, tasks[1].ID); err != nil {
+		t.Fatal(err)
+	}
+
+	counts, err := NewAdminReadRepo(pool).TaskCountsByJobs(ctx, []uuid.UUID{job.ID})
+	if err != nil {
+		t.Fatal(err)
+	}
+	got := counts[job.ID]
+	if got["completed"] != 1 || got["failed"] != 1 || got["pending"] != 1 {
+		t.Errorf("task counts = %v, want completed=1 failed=1 pending=1", got)
+	}
+}
+
+func TestAdminJobCountsByDayAndWorkload(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	repo := NewAdminReadRepo(pool)
+
+	job, _ := seedJob(t, pool, 1)
+	t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM jobs WHERE id = $1`, job.ID) })
+	if _, err := pool.Exec(ctx, `UPDATE jobs SET created_at = now() - interval '2 days' WHERE id = $1`, job.ID); err != nil {
+		t.Fatal(err)
+	}
+	other, _ := seedJob(t, pool, 1)
+	t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM jobs WHERE id = $1`, other.ID) })
+
+	byDay, err := repo.JobCountsByDay(ctx, time.Now().UTC().Add(-6*24*time.Hour))
+	if err != nil {
+		t.Fatal(err)
+	}
+	today := time.Now().UTC().Format("2006-01-02")
+	twoDays := time.Now().UTC().Add(-48 * time.Hour).Format("2006-01-02")
+	if byDay[today] != 1 || byDay[twoDays] != 1 {
+		t.Errorf("by day = %v, want today=1 twoDays=1", byDay)
+	}
+
+	byWorkload, err := repo.JobCountsByWorkload(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if byWorkload[job.Workload] < 2 {
+		t.Errorf("by workload = %v, want at least 2 for %s", byWorkload, job.Workload)
+	}
+}
+
+func TestAdminTaskStatsAndStorage(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	repo := NewAdminReadRepo(pool)
+
+	job, tasks := seedJob(t, pool, 2)
+	t.Cleanup(func() { _, _ = pool.Exec(context.Background(), `DELETE FROM jobs WHERE id = $1`, job.ID) })
+
+	start := time.Now().UTC().Add(-2 * time.Minute)
+	done := time.Now().UTC().Add(-90 * time.Second)
+	resultArtifact := seedArtifact(t, pool, job.ID, &tasks[0].ID, domain.ArtifactPartialResult)
+	if _, err := pool.Exec(ctx, `UPDATE tasks SET status = 'completed', result_artifact_id = $1, started_at = $2, completed_at = $3 WHERE id = $4`,
+		resultArtifact.ID, start, done, tasks[0].ID); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := pool.Exec(ctx, `UPDATE tasks SET status = 'failed' WHERE id = $1`, tasks[1].ID); err != nil {
+		t.Fatal(err)
+	}
+
+	completed, failed, avg, err := repo.TaskStats(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if completed < 1 || failed < 1 {
+		t.Errorf("stats = completed %d failed %d, want >= 1/1", completed, failed)
+	}
+	if avg < 29 || avg > 31 {
+		t.Errorf("avg duration = %.1fs, want ~30s", avg)
+	}
+
+	for _, kind := range []domain.ArtifactKind{domain.ArtifactInput, domain.ArtifactShard} {
+		seedArtifact(t, pool, job.ID, nil, kind)
+	}
+	sizes, err := repo.ArtifactSizeByKind(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if sizes["input"] < 1 || sizes["shard"] < 1 {
+		t.Errorf("sizes = %v, want input/shard > 0", sizes)
+	}
+	dbBytes, err := repo.DatabaseSizeBytes(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if dbBytes <= 0 {
+		t.Errorf("database size = %d, want > 0", dbBytes)
+	}
+}
+
+func TestWorkloadSettingsRepoRoundTrip(t *testing.T) {
+	ensureMigrated(t)
+	pool := testPool(t)
+	ctx := context.Background()
+	repo := NewWorkloadSettingsRepo(pool)
+
+	enabled, err := repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !enabled {
+		t.Error("workload without an override must be enabled")
+	}
+	now := time.Now().UTC()
+	if err := repo.SetEnabled(ctx, "similarity-search", false, now); err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() {
+		_, _ = pool.Exec(context.Background(), `DELETE FROM workload_settings WHERE workload = 'similarity-search'`)
+	})
+	enabled, err = repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if enabled {
+		t.Error("workload must be disabled after the override")
+	}
+	if err := repo.SetEnabled(ctx, "similarity-search", true, now.Add(time.Hour)); err != nil {
+		t.Fatal(err)
+	}
+	enabled, err = repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !enabled {
+		t.Error("workload must be re-enabled after the upsert")
+	}
+	list, err := repo.List(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	found := false
+	for _, s := range list {
+		if s.Workload == "similarity-search" {
+			found = true
+			if !s.Enabled {
+				t.Error("list must reflect the re-enabled state")
+			}
+		}
+	}
+	if !found {
+		t.Error("override missing from the settings list")
+	}
+}
@@ -0,0 +1,198 @@
+package postgres
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	sq "github.com/Masterminds/squirrel"
+	"github.com/google/uuid"
+	"github.com/jackc/pgx/v5/pgxpool"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// AdminReadRepo backs the coordinator admin console: paginated jobs, status
+// counters, metrics buckets and storage figures. Read-only.
+type AdminReadRepo struct{ pool *pgxpool.Pool }
+
+func NewAdminReadRepo(pool *pgxpool.Pool) *AdminReadRepo { return &AdminReadRepo{pool: pool} }
+
+var _ usecase.AdminReadRepository = (*AdminReadRepo)(nil)
+
+func (r *AdminReadRepo) ListJobsPaginated(ctx context.Context, status string, limit, offset int) ([]domain.Job, int, error) {
+	if limit < 1 || limit > 100 || offset < 0 {
+		return nil, 0, domain.ErrInvalidInput
+	}
+	countQ := psql.Select("COUNT(*)").From("jobs")
+	listQ := psql.Select(jobColumns...).From("jobs")
+	if status != "" {
+		countQ = countQ.Where(sq.Eq{"status": status})
+		listQ = listQ.Where(sq.Eq{"status": status})
+	}
+	countSQL, args, err := countQ.ToSql()
+	if err != nil {
+		return nil, 0, err
+	}
+	var total int
+	if err := conn(ctx, r.pool).QueryRow(ctx, countSQL, args...).Scan(&total); err != nil {
+		return nil, 0, fmt.Errorf("count jobs: %w", err)
+	}
+	listSQL, args, err := listQ.OrderBy("created_at DESC", "id DESC").Limit(uint64(limit)).Offset(uint64(offset)).ToSql()
+	if err != nil {
+		return nil, 0, err
+	}
+	rows, err := conn(ctx, r.pool).Query(ctx, listSQL, args...)
+	if err != nil {
+		return nil, 0, fmt.Errorf("list jobs paginated: %w", err)
+	}
+	defer rows.Close()
+	jobs := make([]domain.Job, 0)
+	for rows.Next() {
+		var j domain.Job
+		var statusRaw string
+		if err := rows.Scan(
+			&j.ID, &j.Workload, &j.InputURI, &j.Parameters, &statusRaw, &j.CreatedAt, &j.CompletedAt,
+			&j.InputArtifactID, &j.ResultArtifactID, &j.ErrorCode, &j.ErrorMessage, &j.ReducerStartedAt,
+			&j.OwnerID,
+		); err != nil {
+			return nil, 0, err
+		}
+		j.Status = domain.JobStatus(statusRaw)
+		jobs = append(jobs, j)
+	}
+	return jobs, total, rows.Err()
+}
+
+func (r *AdminReadRepo) CountJobsByStatus(ctx context.Context) (map[string]int, error) {
+	rows, err := conn(ctx, r.pool).Query(ctx, "SELECT status, COUNT(*) FROM jobs GROUP BY status")
+	if err != nil {
+		return nil, fmt.Errorf("count jobs by status: %w", err)
+	}
+	defer rows.Close()
+	out := make(map[string]int)
+	for rows.Next() {
+		var status string
+		var count int
+		if err := rows.Scan(&status, &count); err != nil {
+			return nil, err
+		}
+		out[status] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) TaskCountsByJobs(ctx context.Context, jobIDs []uuid.UUID) (map[uuid.UUID]map[string]int, error) {
+	out := make(map[uuid.UUID]map[string]int, len(jobIDs))
+	if len(jobIDs) == 0 {
+		return out, nil
+	}
+	sql, args, err := psql.Select("job_id", "status", "COUNT(*)").From("tasks").
+		Where(sq.Eq{"job_id": jobIDs}).GroupBy("job_id", "status").ToSql()
+	if err != nil {
+		return nil, err
+	}
+	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
+	if err != nil {
+		return nil, fmt.Errorf("task counts by jobs: %w", err)
+	}
+	defer rows.Close()
+	for rows.Next() {
+		var jobID uuid.UUID
+		var status string
+		var count int
+		if err := rows.Scan(&jobID, &status, &count); err != nil {
+			return nil, err
+		}
+		if out[jobID] == nil {
+			out[jobID] = make(map[string]int)
+		}
+		out[jobID][status] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) JobCountsByDay(ctx context.Context, since time.Time) (map[string]int, error) {
+	rows, err := conn(ctx, r.pool).Query(ctx,
+		"SELECT to_char(date_trunc('day', created_at AT TIME ZONE 'UTC'), 'YYYY-MM-DD') AS day, COUNT(*) FROM jobs WHERE created_at >= $1 GROUP BY 1",
+		since.UTC())
+	if err != nil {
+		return nil, fmt.Errorf("job counts by day: %w", err)
+	}
+	defer rows.Close()
+	out := make(map[string]int)
+	for rows.Next() {
+		var day string
+		var count int
+		if err := rows.Scan(&day, &count); err != nil {
+			return nil, err
+		}
+		out[day] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) JobCountsByWorkload(ctx context.Context) (map[string]int, error) {
+	rows, err := conn(ctx, r.pool).Query(ctx, "SELECT workload, COUNT(*) FROM jobs GROUP BY workload")
+	if err != nil {
+		return nil, fmt.Errorf("job counts by workload: %w", err)
+	}
+	defer rows.Close()
+	out := make(map[string]int)
+	for rows.Next() {
+		var workload string
+		var count int
+		if err := rows.Scan(&workload, &count); err != nil {
+			return nil, err
+		}
+		out[workload] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) TaskStats(ctx context.Context) (int64, int64, float64, error) {
+	var completed, failed int64
+	var avgSeconds *float64
+	err := conn(ctx, r.pool).QueryRow(ctx, `
+		SELECT
+			COALESCE(SUM(CASE WHEN status = 'completed' THEN 1 ELSE 0 END), 0),
+			COALESCE(SUM(CASE WHEN status = 'failed' THEN 1 ELSE 0 END), 0),
+			AVG(CASE WHEN status = 'completed' AND started_at IS NOT NULL
+				THEN EXTRACT(EPOCH FROM completed_at - started_at) END)::float8
+		FROM tasks`).Scan(&completed, &failed, &avgSeconds)
+	if err != nil {
+		return 0, 0, 0, fmt.Errorf("task stats: %w", err)
+	}
+	var avg float64
+	if avgSeconds != nil {
+		avg = *avgSeconds
+	}
+	return completed, failed, avg, nil
+}
+
+func (r *AdminReadRepo) ArtifactSizeByKind(ctx context.Context) (map[string]int64, error) {
+	rows, err := conn(ctx, r.pool).Query(ctx, "SELECT kind, COALESCE(SUM(size_bytes), 0) FROM artifacts GROUP BY kind")
+	if err != nil {
+		return nil, fmt.Errorf("artifact sizes: %w", err)
+	}
+	defer rows.Close()
+	out := make(map[string]int64)
+	for rows.Next() {
+		var kind string
+		var size int64
+		if err := rows.Scan(&kind, &size); err != nil {
+			return nil, err
+		}
+		out[kind] = size
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) DatabaseSizeBytes(ctx context.Context) (int64, error) {
+	var size int64
+	if err := conn(ctx, r.pool).QueryRow(ctx, "SELECT pg_database_size(current_database())").Scan(&size); err != nil {
+		return 0, fmt.Errorf("database size: %w", err)
+	}
+	return size, nil
+}
@@ -24,6 +24,7 @@ import (

 	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
 	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+	"github.com/emil28092005/SciMesh/coordinator/internal/workloads"
 )

 func testPool(t *testing.T) *pgxpool.Pool {
@@ -144,20 +145,20 @@ func TestUIReadRepoListsReducerFields(t *testing.T) {
 	if claimed, err := jobs.ClaimReduction(ctx, job.ID, time.Now().UTC()); err != nil || !claimed {
 		t.Fatalf("claim reduction = (%v, %v)", claimed, err)
 	}
-	listed, err := NewUIReadRepo(pool).ListJobs(ctx, nil, 20)
+	listed, _, err := NewAdminReadRepo(pool).ListJobsPaginated(ctx, "", 20, 0)
 	if err != nil {
-		t.Fatalf("list UI jobs: %v", err)
+		t.Fatalf("list admin jobs: %v", err)
 	}
 	for _, item := range listed {
 		if item.ID != job.ID {
 			continue
 		}
 		if item.Status != domain.JobReducing || item.ReducerStartedAt == nil {
-			t.Fatalf("UI reducer projection = %+v", item)
+			t.Fatalf("admin reducer projection = %+v", item)
 		}
 		return
 	}
-	t.Fatalf("seeded job %s is missing from UI list", job.ID)
+	t.Fatalf("seeded job %s is missing from the admin list", job.ID)
 }

 // A job must land whole or not at all: a half-created job leaves chunks no
@@ -408,7 +409,7 @@ func TestCompleteTaskReplayIsIdempotent(t *testing.T) {
 	tasks, jobs, artifacts, tx := NewTaskRepo(pool), NewJobRepo(pool), NewArtifactRepo(pool), NewTxManager(pool)
 	workers, results := NewWorkerRepo(pool), NewTaskResultRepo(pool)
 	clk := fixedClock{now: time.Now().UTC()}
-	uc := usecase.NewCompleteTask(tasks, jobs, artifacts, workers, results, tx, clk, 2)
+	uc := usecase.NewCompleteTask(tasks, jobs, artifacts, workers, results, tx, clk, 2, integrationCatalog())

 	claimed, err := tasks.ClaimNext(ctx, usecase.ClaimFilter{
 		Owner: "worker-1", Now: clk.now, LeaseUntil: clk.now.Add(time.Minute),
@@ -657,3 +658,46 @@ func TestExpireLeasesRequeuesElapsedTasks(t *testing.T) {
 		t.Errorf("pending = %d, want 1 — a dead worker must not strand its task", counts[domain.TaskPending])
 	}
 }
+
+func TestMigrateProvisionsAndIsIdempotent(t *testing.T) {
+	ctx := context.Background()
+	url := os.Getenv("TEST_DATABASE_URL")
+	if url == "" {
+		t.Skip("TEST_DATABASE_URL is not set")
+	}
+	if err := Migrate(ctx, url, nil); err != nil {
+		t.Fatalf("first migrate: %v", err)
+	}
+	if err := Migrate(ctx, url, nil); err != nil {
+		t.Fatalf("second migrate (idempotent): %v", err)
+	}
+	pool := testPool(t)
+	migrations, err := listMigrations()
+	if err != nil {
+		t.Fatal(err)
+	}
+	var watermark int64
+	if err := pool.QueryRow(ctx, "SELECT COALESCE(MAX(version), 0) FROM schema_migrations").Scan(&watermark); err != nil {
+		t.Fatalf("read schema_migrations: %v", err)
+	}
+	if watermark != int64(len(migrations)) {
+		t.Errorf("schema watermark = %d, want %d", watermark, len(migrations))
+	}
+	var hasJobs bool
+	if err := pool.QueryRow(ctx,
+		"SELECT EXISTS (SELECT 1 FROM information_schema.tables WHERE table_name = 'jobs')",
+	).Scan(&hasJobs); err != nil {
+		t.Fatal(err)
+	}
+	if !hasJobs {
+		t.Error("jobs table was not created by the embedded migrations")
+	}
+}
+
+func integrationCatalog() *workloads.Catalog {
+	catalog, err := workloads.Load()
+	if err != nil {
+		panic(err)
+	}
+	return catalog
+}
@@ -3,6 +3,7 @@ package postgres
 import (
 	"context"
 	"errors"
+	"fmt"
 	"time"

 	sq "github.com/Masterminds/squirrel"
@@ -153,3 +154,46 @@ func (r *JobRepo) UpdateStatus(ctx context.Context, id uuid.UUID,
 	}
 	return nil
 }
+
+// ListCompletedBefore returns jobs whose completion timestamp is older than
+// the cutoff (completed and failed both count as finished).
+func (r *JobRepo) ListCompletedBefore(ctx context.Context, cutoff time.Time) ([]domain.Job, error) {
+	sql, args, err := psql.Select(jobColumns...).From("jobs").
+		Where(sq.NotEq{"completed_at": nil}).
+		Where(sq.Lt{"completed_at": cutoff}).
+		OrderBy("completed_at ASC").
+		ToSql()
+	if err != nil {
+		return nil, err
+	}
+	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
+	if err != nil {
+		return nil, fmt.Errorf("list completed jobs: %w", err)
+	}
+	defer rows.Close()
+	var jobs []domain.Job
+	for rows.Next() {
+		var j domain.Job
+		var status string
+		if err := rows.Scan(
+			&j.ID, &j.Workload, &j.InputURI, &j.Parameters, &status, &j.CreatedAt, &j.CompletedAt,
+			&j.InputArtifactID, &j.ResultArtifactID, &j.ErrorCode, &j.ErrorMessage, &j.ReducerStartedAt,
+			&j.OwnerID,
+		); err != nil {
+			return nil, err
+		}
+		j.Status = domain.JobStatus(status)
+		jobs = append(jobs, j)
+	}
+	return jobs, rows.Err()
+}
+
+// Delete removes the job row; tasks, artifacts and task_results cascade.
+func (r *JobRepo) Delete(ctx context.Context, id uuid.UUID) error {
+	sql, args, err := psql.Delete("jobs").Where(sq.Eq{"id": id}).ToSql()
+	if err != nil {
+		return err
+	}
+	_, err = conn(ctx, r.pool).Exec(ctx, sql, args...)
+	return err
+}
@@ -0,0 +1,144 @@
+package postgres
+
+import (
+	"context"
+	"embed"
+	"fmt"
+	"log/slog"
+	"regexp"
+	"sort"
+	"strconv"
+
+	"github.com/jackc/pgx/v5"
+)
+
+//go:embed migrations/*.sql
+var migrationFiles embed.FS
+
+var migrationNamePattern = regexp.MustCompile(`^([0-9]+)_[a-z0-9_]+\.(up|down)\.sql$`)
+
+// migration is one parsed embedded migration file.
+type migration struct {
+	version int
+	name    string
+	sql     string
+}
+
+// listMigrations parses and orders the embedded .up.sql files by version.
+func listMigrations() ([]migration, error) {
+	entries, err := migrationFiles.ReadDir("migrations")
+	if err != nil {
+		return nil, fmt.Errorf("read embedded migrations: %w", err)
+	}
+	up := map[int]migration{}
+	for _, entry := range entries {
+		match := migrationNamePattern.FindStringSubmatch(entry.Name())
+		if match == nil {
+			continue
+		}
+		if match[2] != "up" {
+			continue
+		}
+		version, err := strconv.Atoi(match[1])
+		if err != nil {
+			return nil, fmt.Errorf("migration %q has an invalid version: %w", entry.Name(), err)
+		}
+		if _, duplicate := up[version]; duplicate {
+			return nil, fmt.Errorf("migration version %d is duplicated", version)
+		}
+		body, err := migrationFiles.ReadFile("migrations/" + entry.Name())
+		if err != nil {
+			return nil, fmt.Errorf("read migration %q: %w", entry.Name(), err)
+		}
+		up[version] = migration{version: version, name: entry.Name(), sql: string(body)}
+	}
+	if len(up) == 0 {
+		return nil, fmt.Errorf("no .up.sql migrations are embedded")
+	}
+	versions := make([]int, 0, len(up))
+	for version := range up {
+		versions = append(versions, version)
+	}
+	sort.Ints(versions)
+	migrations := make([]migration, 0, len(versions))
+	for _, version := range versions {
+		migrations = append(migrations, up[version])
+	}
+	for index, item := range migrations {
+		if item.version != index+1 {
+			return nil, fmt.Errorf("embedded migrations are not contiguous: version %d at position %d", item.version, index+1)
+		}
+	}
+	return migrations, nil
+}
+
+// Migrate applies every embedded migration above the recorded schema version,
+// so the binary provisions its own schema. It is idempotent and interoperates
+// with the golang-migrate CLI: both tools use the same schema_migrations
+// watermark table (single row: version + dirty flag), and a PostgreSQL
+// advisory lock serializes concurrent migrators. Each migration file runs as
+// its own transaction (the files carry explicit BEGIN/COMMIT, matching the
+// golang-migrate format the CLI and CI still use).
+func Migrate(ctx context.Context, databaseURL string, log *slog.Logger) error {
+	migrations, err := listMigrations()
+	if err != nil {
+		return err
+	}
+	connConfig, err := pgx.ParseConfig(databaseURL)
+	if err != nil {
+		return fmt.Errorf("parse database url: %w", err)
+	}
+	// Migration files contain multiple statements (BEGIN...COMMIT), which the
+	// extended query protocol rejects; run them with the simple protocol.
+	connConfig.DefaultQueryExecMode = pgx.QueryExecModeSimpleProtocol
+	conn, err := pgx.ConnectConfig(ctx, connConfig)
+	if err != nil {
+		return fmt.Errorf("connect for migration: %w", err)
+	}
+	defer func() { _ = conn.Close(ctx) }()
+
+	if _, err := conn.Exec(ctx, "SELECT pg_advisory_lock(82473911)"); err != nil {
+		return fmt.Errorf("acquire migration lock: %w", err)
+	}
+	defer func() { _, _ = conn.Exec(ctx, "SELECT pg_advisory_unlock(82473911)") }()
+
+	if _, err := conn.Exec(ctx,
+		"CREATE TABLE IF NOT EXISTS schema_migrations (version bigint PRIMARY KEY, dirty boolean NOT NULL DEFAULT false)",
+	); err != nil {
+		return fmt.Errorf("ensure schema_migrations: %w", err)
+	}
+
+	var applied int64
+	if err := conn.QueryRow(ctx,
+		"SELECT COALESCE(MAX(version), 0) FROM schema_migrations",
+	).Scan(&applied); err != nil {
+		return fmt.Errorf("read applied schema version: %w", err)
+	}
+
+	for _, item := range migrations {
+		if int64(item.version) <= applied {
+			continue
+		}
+		if log != nil {
+			log.Info("applying migration", "version", item.version, "file", item.name)
+		}
+		if _, err := conn.Exec(ctx, item.sql); err != nil {
+			return fmt.Errorf("apply migration %s: %w", item.name, err)
+		}
+		// Advance the watermark to the single-row golang-migrate layout.
+		tag, err := conn.Exec(ctx,
+			"UPDATE schema_migrations SET version = $1, dirty = false", item.version)
+		if err != nil {
+			return fmt.Errorf("record migration %s: %w", item.name, err)
+		}
+		if tag.RowsAffected() == 0 {
+			if _, err := conn.Exec(ctx,
+				"INSERT INTO schema_migrations (version, dirty) VALUES ($1, false)",
+				item.version,
+			); err != nil {
+				return fmt.Errorf("record migration %s: %w", item.name, err)
+			}
+		}
+	}
+	return nil
+}
@@ -0,0 +1,62 @@
+package postgres
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestListMigrationsParsesAndOrdersEmbeddedFiles(t *testing.T) {
+	migrations, err := listMigrations()
+	if err != nil {
+		t.Fatalf("list migrations: %v", err)
+	}
+	if len(migrations) == 0 {
+		t.Fatal("no embedded migrations")
+	}
+	for index, item := range migrations {
+		if item.version != index+1 {
+			t.Errorf("migration %d has version %d, want contiguous ordering", index, item.version)
+		}
+		if item.name != expectedMigrationName(item.version) {
+			t.Errorf("migration %d file is %q, want %q", item.version, item.name, expectedMigrationName(item.version))
+		}
+		if strings.TrimSpace(item.sql) == "" {
+			t.Errorf("migration %d is empty", item.version)
+		}
+	}
+}
+
+func expectedMigrationName(version int) string {
+	switch version {
+	case 1:
+		return "0001_init.up.sql"
+	case 2:
+		return "0002_workers.up.sql"
+	case 3:
+		return "0003_artifacts.up.sql"
+	case 4:
+		return "0004_result_artifact.up.sql"
+	case 5:
+		return "0005_uploaded_input.up.sql"
+	case 6:
+		return "0006_task_running_enum.up.sql"
+	case 7:
+		return "0007_task_running_lease.up.sql"
+	case 8:
+		return "0008_artifact_attempt.up.sql"
+	case 9:
+		return "0009_unique_partial_result_attempt.up.sql"
+	case 10:
+		return "0010_job_reduction.up.sql"
+	case 11:
+		return "0011_job_owner.up.sql"
+	case 12:
+		return "0012_worker_trust.up.sql"
+	case 13:
+		return "0013_task_results.up.sql"
+	case 14:
+		return "0014_workload_settings.up.sql"
+	default:
+		return ""
+	}
+}
@@ -0,0 +1,5 @@
+BEGIN;
+
+DROP TABLE IF EXISTS workload_settings;
+
+COMMIT;
@@ -0,0 +1,11 @@
+BEGIN;
+
+-- Per-workload enable/disable. Absence of a row means "enabled" (the catalog
+-- default); a row only exists once an admin flipped a workload off or back on.
+CREATE TABLE workload_settings (
+    workload   text        NOT NULL PRIMARY KEY,
+    enabled    boolean     NOT NULL,
+    updated_at timestamptz NOT NULL DEFAULT now()
+);
+
+COMMIT;
@@ -24,40 +24,6 @@ func (r *UIReadRepo) GetJob(ctx context.Context, id uuid.UUID) (*domain.Job, err
 	return job, err
 }

-func (r *UIReadRepo) ListJobs(ctx context.Context, owner *uuid.UUID, limit int) ([]domain.Job, error) {
-	if limit < 1 || limit > 100 {
-		return nil, domain.ErrInvalidInput
-	}
-	q := psql.Select(jobColumns...).From("jobs")
-	if owner != nil {
-		q = q.Where(sq.Eq{"owner_id": *owner})
-	}
-	sql, args, err := q.OrderBy("created_at DESC", "id DESC").Limit(uint64(limit)).ToSql()
-	if err != nil {
-		return nil, err
-	}
-	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
-	if err != nil {
-		return nil, fmt.Errorf("list jobs: %w", err)
-	}
-	defer rows.Close()
-	jobs := make([]domain.Job, 0)
-	for rows.Next() {
-		var j domain.Job
-		var status string
-		if err := rows.Scan(
-			&j.ID, &j.Workload, &j.InputURI, &j.Parameters, &status, &j.CreatedAt, &j.CompletedAt,
-			&j.InputArtifactID, &j.ResultArtifactID, &j.ErrorCode, &j.ErrorMessage, &j.ReducerStartedAt,
-			&j.OwnerID,
-		); err != nil {
-			return nil, err
-		}
-		j.Status = domain.JobStatus(status)
-		jobs = append(jobs, j)
-	}
-	return jobs, rows.Err()
-}
-
 func (r *UIReadRepo) ListTasksByJob(ctx context.Context, jobID uuid.UUID) ([]domain.Task, error) {
 	sql, args, err := psql.Select(taskColumns...).From("tasks").Where(sq.Eq{"job_id": jobID}).OrderBy("chunk_index ASC").ToSql()
 	if err != nil {
@@ -79,68 +45,18 @@ func (r *UIReadRepo) ListTasksByJob(ctx context.Context, jobID uuid.UUID) ([]dom
 	return tasks, rows.Err()
 }

-func (r *UIReadRepo) ListTasksByJobs(ctx context.Context, jobIDs []uuid.UUID) (map[uuid.UUID][]domain.Task, error) {
-	out := make(map[uuid.UUID][]domain.Task, len(jobIDs))
-	if len(jobIDs) == 0 {
-		return out, nil
-	}
-	sql, args, err := psql.Select(taskColumns...).From("tasks").
-		Where(sq.Eq{"job_id": jobIDs}).OrderBy("job_id ASC", "chunk_index ASC").ToSql()
-	if err != nil {
-		return nil, err
-	}
-	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
-	if err != nil {
-		return nil, fmt.Errorf("list tasks by jobs: %w", err)
-	}
-	defer rows.Close()
-	for rows.Next() {
-		task, err := scanTask(rows)
-		if err != nil {
-			return nil, err
-		}
-		out[task.JobID] = append(out[task.JobID], *task)
-	}
-	return out, rows.Err()
-}
-
 func (r *UIReadRepo) ListWorkers(ctx context.Context, limit int) ([]domain.Worker, error) {
-	if limit < 1 || limit > 100 {
-		return nil, domain.ErrInvalidInput
-	}
-	sql, args, err := psql.Select(workerColumns...).From("workers").OrderBy("last_heartbeat_at DESC", "id DESC").Limit(uint64(limit)).ToSql()
-	if err != nil {
-		return nil, err
-	}
-	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
-	if err != nil {
-		return nil, fmt.Errorf("list workers: %w", err)
-	}
-	defer rows.Close()
-	workers := make([]domain.Worker, 0)
-	for rows.Next() {
-		worker, err := scanWorker(rows)
-		if err != nil {
-			return nil, err
-		}
-		workers = append(workers, *worker)
-	}
-	return workers, rows.Err()
-}
-
-func (r *UIReadRepo) ListWorkersByOwner(ctx context.Context, owner uuid.UUID, limit int) ([]domain.Worker, error) {
 	if limit < 1 || limit > 100 {
 		return nil, domain.ErrInvalidInput
 	}
 	sql, args, err := psql.Select(workerColumns...).From("workers").
-		Where(sq.Eq{"owner_id": owner}).
 		OrderBy("last_heartbeat_at DESC", "id DESC").Limit(uint64(limit)).ToSql()
 	if err != nil {
 		return nil, err
 	}
 	rows, err := conn(ctx, r.pool).Query(ctx, sql, args...)
 	if err != nil {
-		return nil, fmt.Errorf("list workers by owner: %w", err)
+		return nil, fmt.Errorf("list workers: %w", err)
 	}
 	defer rows.Close()
 	workers := make([]domain.Worker, 0)
@@ -91,6 +91,24 @@ func (r *WorkerRepo) MarkStaleOffline(ctx context.Context, cutoff time.Time) (in
 	return tag.RowsAffected(), nil
 }

+func (r *WorkerRepo) SetTrust(ctx context.Context, id uuid.UUID, trust domain.WorkerTrust) error {
+	sql, args, err := psql.Update("workers").
+		SetMap(map[string]any{"trust_level": string(trust), "updated_at": time.Now()}).
+		Where(sq.Eq{"id": id}).
+		ToSql()
+	if err != nil {
+		return err
+	}
+	tag, err := conn(ctx, r.pool).Exec(ctx, sql, args...)
+	if err != nil {
+		return fmt.Errorf("set worker trust: %w", err)
+	}
+	if tag.RowsAffected() == 0 {
+		return domain.ErrWorkerNotFound
+	}
+	return nil
+}
+
 func scanWorker(row pgx.Row) (*domain.Worker, error) {
 	var (
 		w      domain.Worker
@@ -105,3 +123,19 @@ func scanWorker(row pgx.Row) (*domain.Worker, error) {
 	w.TrustLevel = domain.WorkerTrust(trust)
 	return &w, nil
 }
+
+// Delete removes a worker from the registry.
+func (r *WorkerRepo) Delete(ctx context.Context, id uuid.UUID) error {
+	sql, args, err := psql.Delete("workers").Where(sq.Eq{"id": id}).ToSql()
+	if err != nil {
+		return err
+	}
+	tag, err := conn(ctx, r.pool).Exec(ctx, sql, args...)
+	if err != nil {
+		return fmt.Errorf("delete worker: %w", err)
+	}
+	if tag.RowsAffected() == 0 {
+		return domain.ErrWorkerNotFound
+	}
+	return nil
+}
@@ -0,0 +1,67 @@
+package postgres
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// WorkloadSettingsRepo persists the per-workload enable/disable overrides.
+// Absence of a row means the workload is enabled (the catalog default).
+type WorkloadSettingsRepo struct{ pool *pgxpool.Pool }
+
+func NewWorkloadSettingsRepo(pool *pgxpool.Pool) *WorkloadSettingsRepo {
+	return &WorkloadSettingsRepo{pool: pool}
+}
+
+var _ usecase.WorkloadSettingsRepository = (*WorkloadSettingsRepo)(nil)
+
+func (r *WorkloadSettingsRepo) GetEnabled(ctx context.Context, workload string) (bool, error) {
+	var enabled bool
+	err := conn(ctx, r.pool).QueryRow(ctx,
+		"SELECT enabled FROM workload_settings WHERE workload = $1", workload).Scan(&enabled)
+	if err != nil && err.Error() == "no rows in result set" {
+		return true, nil // no override: catalog default enabled
+	}
+	if err != nil {
+		return false, fmt.Errorf("get workload setting: %w", err)
+	}
+	return enabled, nil
+}
+
+func (r *WorkloadSettingsRepo) List(ctx context.Context) ([]usecase.WorkloadSetting, error) {
+	rows, err := conn(ctx, r.pool).Query(ctx,
+		"SELECT workload, enabled, updated_at FROM workload_settings ORDER BY workload ASC")
+	if err != nil {
+		return nil, fmt.Errorf("list workload settings: %w", err)
+	}
+	defer rows.Close()
+	var out []usecase.WorkloadSetting
+	for rows.Next() {
+		var s usecase.WorkloadSetting
+		if err := rows.Scan(&s.Workload, &s.Enabled, &s.UpdatedAt); err != nil {
+			return nil, err
+		}
+		out = append(out, s)
+	}
+	return out, rows.Err()
+}
+
+func (r *WorkloadSettingsRepo) SetEnabled(ctx context.Context, workload string, enabled bool, now time.Time) error {
+	sql, args, err := psql.Insert("workload_settings").
+		Columns("workload", "enabled", "updated_at").
+		Values(workload, enabled, now).
+		Suffix(`ON CONFLICT (workload) DO UPDATE SET enabled = EXCLUDED.enabled, updated_at = EXCLUDED.updated_at`).
+		ToSql()
+	if err != nil {
+		return err
+	}
+	if _, err := conn(ctx, r.pool).Exec(ctx, sql, args...); err != nil {
+		return fmt.Errorf("set workload setting: %w", err)
+	}
+	return nil
+}
@@ -0,0 +1,134 @@
+package sqlite
+
+import (
+	"context"
+	"errors"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+func TestWorkloadSettingsRepoRoundTrip(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	repo := NewWorkloadSettingsRepo(db)
+
+	// No override: enabled by default.
+	enabled, err := repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !enabled {
+		t.Error("workload without an override must be enabled")
+	}
+
+	now := time.Date(2026, 8, 2, 12, 0, 0, 0, time.UTC)
+	if err := repo.SetEnabled(ctx, "similarity-search", false, now); err != nil {
+		t.Fatal(err)
+	}
+	enabled, err = repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if enabled {
+		t.Error("workload must be disabled after the override")
+	}
+
+	// Upsert flips it back and updates the timestamp.
+	later := now.Add(time.Hour)
+	if err := repo.SetEnabled(ctx, "similarity-search", true, later); err != nil {
+		t.Fatal(err)
+	}
+	enabled, err = repo.GetEnabled(ctx, "similarity-search")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !enabled {
+		t.Error("workload must be re-enabled after the upsert")
+	}
+
+	list, err := repo.List(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(list) != 1 || list[0].Workload != "similarity-search" || !list[0].Enabled {
+		t.Errorf("list = %+v, want the single re-enabled override", list)
+	}
+}
+
+func TestWorkerSetTrust(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	repo := NewWorkerRepo(db)
+
+	worker, err := domain.NewWorker("lab-node", []string{"similarity-search"}, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.Insert(ctx, worker); err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.SetTrust(ctx, worker.ID, domain.WorkerUntrusted); err != nil {
+		t.Fatal(err)
+	}
+	got, err := repo.Get(ctx, worker.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.TrustLevel != domain.WorkerUntrusted {
+		t.Errorf("trust = %q, want untrusted", got.TrustLevel)
+	}
+	if err := repo.SetTrust(ctx, worker.ID, domain.WorkerTrusted); err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.SetTrust(ctx, uuid.New(), domain.WorkerTrusted); !errors.Is(err, domain.ErrWorkerNotFound) {
+		t.Errorf("unknown worker trust err = %v, want ErrWorkerNotFound", err)
+	}
+}
+
+func TestJobRepoListCompletedBeforeAndDelete(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	repo := NewJobRepo(db)
+
+	old := seedJob(t, db, 2)
+	oldTime := fixedTime().Add(-40 * 24 * time.Hour)
+	if err := repo.UpdateStatus(ctx, old.ID, domain.JobCompleted, &oldTime); err != nil {
+		t.Fatal(err)
+	}
+	fresh := seedJob(t, db, 2)
+	freshTime := fixedTime().Add(-2 * time.Hour)
+	if err := repo.UpdateStatus(ctx, fresh.ID, domain.JobCompleted, &freshTime); err != nil {
+		t.Fatal(err)
+	}
+	// The failing check constraint needs no result artifact for completed; the
+	// UpdateStatus path is fine, but tasks stay pending — irrelevant here.
+
+	list, err := repo.ListCompletedBefore(ctx, fixedTime().Add(-7*24*time.Hour))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(list) != 1 || list[0].ID != old.ID {
+		t.Errorf("list = %d jobs, want only the old one", len(list))
+	}
+	if err := repo.Delete(ctx, old.ID); err != nil {
+		t.Fatal(err)
+	}
+	var n int
+	if err := db.QueryRowContext(ctx, "SELECT COUNT(*) FROM jobs WHERE id = ?", old.ID.String()).Scan(&n); err != nil {
+		t.Fatal(err)
+	}
+	if n != 0 {
+		t.Error("job row must be gone after Delete")
+	}
+	// Tasks cascaded away with the job.
+	if err := db.QueryRowContext(ctx, "SELECT COUNT(*) FROM tasks WHERE job_id = ?", old.ID.String()).Scan(&n); err != nil {
+		t.Fatal(err)
+	}
+	if n != 0 {
+		t.Error("tasks must cascade with the job")
+	}
+}
@@ -0,0 +1,191 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// AdminReadRepo backs the coordinator admin console: paginated jobs, status
+// counters, metrics buckets and storage figures. Read-only.
+type AdminReadRepo struct{ db *sql.DB }
+
+func NewAdminReadRepo(db *sql.DB) *AdminReadRepo { return &AdminReadRepo{db: db} }
+
+var _ usecase.AdminReadRepository = (*AdminReadRepo)(nil)
+
+func (r *AdminReadRepo) ListJobsPaginated(ctx context.Context, status string, limit, offset int) ([]domain.Job, int, error) {
+	if limit < 1 || limit > 100 || offset < 0 {
+		return nil, 0, domain.ErrInvalidInput
+	}
+	where := ""
+	args := []any{}
+	if status != "" {
+		where = " WHERE status = ?"
+		args = append(args, status)
+	}
+	var total int
+	if err := conn(ctx, r.db).QueryRowContext(ctx, "SELECT COUNT(*) FROM jobs"+where, args...).Scan(&total); err != nil {
+		return nil, 0, fmt.Errorf("count jobs: %w", err)
+	}
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+jobColumns+" FROM jobs"+where+" ORDER BY created_at DESC, id DESC LIMIT ? OFFSET ?",
+		append(args, limit, offset)...)
+	if err != nil {
+		return nil, 0, fmt.Errorf("list jobs paginated: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	jobs := make([]domain.Job, 0)
+	for rows.Next() {
+		job, err := scanJob(rows)
+		if err != nil {
+			return nil, 0, err
+		}
+		jobs = append(jobs, *job)
+	}
+	return jobs, total, rows.Err()
+}
+
+func (r *AdminReadRepo) CountJobsByStatus(ctx context.Context) (map[string]int, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx, "SELECT status, COUNT(*) FROM jobs GROUP BY status")
+	if err != nil {
+		return nil, fmt.Errorf("count jobs by status: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	out := make(map[string]int)
+	for rows.Next() {
+		var status string
+		var count int
+		if err := rows.Scan(&status, &count); err != nil {
+			return nil, err
+		}
+		out[status] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) TaskCountsByJobs(ctx context.Context, jobIDs []uuid.UUID) (map[uuid.UUID]map[string]int, error) {
+	out := make(map[uuid.UUID]map[string]int, len(jobIDs))
+	if len(jobIDs) == 0 {
+		return out, nil
+	}
+	placeholders := make([]string, 0, len(jobIDs))
+	args := make([]any, 0, len(jobIDs))
+	for _, id := range jobIDs {
+		placeholders = append(placeholders, "?")
+		args = append(args, id.String())
+	}
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT job_id, status, COUNT(*) FROM tasks WHERE job_id IN ("+strings.Join(placeholders, ", ")+") GROUP BY job_id, status",
+		args...)
+	if err != nil {
+		return nil, fmt.Errorf("task counts by jobs: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	for rows.Next() {
+		var jobRaw, status string
+		var count int
+		if err := rows.Scan(&jobRaw, &status, &count); err != nil {
+			return nil, err
+		}
+		jobID, err := uuid.Parse(jobRaw)
+		if err != nil {
+			return nil, fmt.Errorf("task counts: parse job id: %w", err)
+		}
+		if out[jobID] == nil {
+			out[jobID] = make(map[string]int)
+		}
+		out[jobID][status] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) JobCountsByDay(ctx context.Context, since time.Time) (map[string]int, error) {
+	// created_at is unix nanos; the bucket is the UTC calendar day.
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT strftime('%Y-%m-%d', created_at / 1000000000, 'unixepoch') AS day, COUNT(*) FROM jobs WHERE created_at >= ? GROUP BY day",
+		since.UTC().UnixNano())
+	if err != nil {
+		return nil, fmt.Errorf("job counts by day: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	out := make(map[string]int)
+	for rows.Next() {
+		var day string
+		var count int
+		if err := rows.Scan(&day, &count); err != nil {
+			return nil, err
+		}
+		out[day] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) JobCountsByWorkload(ctx context.Context) (map[string]int, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx, "SELECT workload, COUNT(*) FROM jobs GROUP BY workload")
+	if err != nil {
+		return nil, fmt.Errorf("job counts by workload: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	out := make(map[string]int)
+	for rows.Next() {
+		var workload string
+		var count int
+		if err := rows.Scan(&workload, &count); err != nil {
+			return nil, err
+		}
+		out[workload] = count
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) TaskStats(ctx context.Context) (int64, int64, float64, error) {
+	var completed, failed int64
+	var avgNanos sql.NullFloat64
+	err := conn(ctx, r.db).QueryRowContext(ctx, `
+		SELECT
+			COALESCE(SUM(CASE WHEN status = 'completed' THEN 1 ELSE 0 END), 0),
+			COALESCE(SUM(CASE WHEN status = 'failed' THEN 1 ELSE 0 END), 0),
+			AVG(CASE WHEN status = 'completed' AND started_at IS NOT NULL THEN completed_at - started_at END)
+		FROM tasks`).Scan(&completed, &failed, &avgNanos)
+	if err != nil {
+		return 0, 0, 0, fmt.Errorf("task stats: %w", err)
+	}
+	return completed, failed, avgNanos.Float64 / 1e9, nil
+}
+
+func (r *AdminReadRepo) ArtifactSizeByKind(ctx context.Context) (map[string]int64, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx, "SELECT kind, COALESCE(SUM(size_bytes), 0) FROM artifacts GROUP BY kind")
+	if err != nil {
+		return nil, fmt.Errorf("artifact sizes: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	out := make(map[string]int64)
+	for rows.Next() {
+		var kind string
+		var size int64
+		if err := rows.Scan(&kind, &size); err != nil {
+			return nil, err
+		}
+		out[kind] = size
+	}
+	return out, rows.Err()
+}
+
+func (r *AdminReadRepo) DatabaseSizeBytes(ctx context.Context) (int64, error) {
+	var pageCount, pageSize int64
+	if err := conn(ctx, r.db).QueryRowContext(ctx, "PRAGMA page_count").Scan(&pageCount); err != nil {
+		return 0, fmt.Errorf("page count: %w", err)
+	}
+	if err := conn(ctx, r.db).QueryRowContext(ctx, "PRAGMA page_size").Scan(&pageSize); err != nil {
+		return 0, fmt.Errorf("page size: %w", err)
+	}
+	return pageCount * pageSize, nil
+}
@@ -0,0 +1,171 @@
+package sqlite
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+func TestAdminListJobsPaginatedAndCounts(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	jobRepo := NewJobRepo(db)
+	adminRepo := NewAdminReadRepo(db)
+
+	jobs := make([]*domain.Job, 5)
+	for i := range jobs {
+		jobs[i] = seedJob(t, db, 2)
+	}
+	// Two completed, two running, one pending.
+	if err := jobRepo.UpdateStatus(ctx, jobs[0].ID, domain.JobCompleted, nil); err != nil {
+		t.Fatal(err)
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[1].ID, domain.JobCompleted, nil); err != nil {
+		t.Fatal(err)
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[2].ID, domain.JobRunning, nil); err != nil {
+		t.Fatal(err)
+	}
+	if err := jobRepo.UpdateStatus(ctx, jobs[3].ID, domain.JobRunning, nil); err != nil {
+		t.Fatal(err)
+	}
+
+	all, total, err := adminRepo.ListJobsPaginated(ctx, "", 100, 0)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 5 || len(all) != 5 {
+		t.Errorf("all: total=%d len=%d, want 5/5", total, len(all))
+	}
+	completed, total, err := adminRepo.ListJobsPaginated(ctx, "completed", 100, 0)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 2 || len(completed) != 2 {
+		t.Errorf("completed: total=%d len=%d, want 2/2", total, len(completed))
+	}
+	page, total, err := adminRepo.ListJobsPaginated(ctx, "", 2, 2)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if total != 5 || len(page) != 2 {
+		t.Errorf("page: total=%d len=%d, want 5/2", total, len(page))
+	}
+
+	counts, err := adminRepo.CountJobsByStatus(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if counts["completed"] != 2 || counts["running"] != 2 || counts["pending"] != 1 {
+		t.Errorf("counts = %v, want completed=2 running=2 pending=1", counts)
+	}
+}
+
+func TestAdminTaskCountsByJobs(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 3)
+	if _, err := db.ExecContext(ctx, "UPDATE tasks SET status = 'completed', result_artifact_id = ? WHERE chunk_index = 0 AND job_id = ?", uuid.NewString(), job.ID.String()); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := db.ExecContext(ctx, "UPDATE tasks SET status = 'failed' WHERE chunk_index = 1 AND job_id = ?", job.ID.String()); err != nil {
+		t.Fatal(err)
+	}
+
+	counts, err := NewAdminReadRepo(db).TaskCountsByJobs(ctx, []uuid.UUID{job.ID})
+	if err != nil {
+		t.Fatal(err)
+	}
+	got := counts[job.ID]
+	if got["completed"] != 1 || got["failed"] != 1 || got["pending"] != 1 {
+		t.Errorf("task counts = %v, want completed=1 failed=1 pending=1", got)
+	}
+}
+
+func TestAdminJobCountsByDay(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 1)
+	// Move the seed job to two days ago; create two more today.
+	old := fixedTime().Add(-48 * time.Hour)
+	if _, err := db.ExecContext(ctx, "UPDATE jobs SET created_at = ? WHERE id = ?", old.UnixNano(), job.ID.String()); err != nil {
+		t.Fatal(err)
+	}
+	seedJob(t, db, 1)
+	seedJob(t, db, 1)
+
+	repo := NewAdminReadRepo(db)
+	counts, err := repo.JobCountsByDay(ctx, fixedTime().Add(-6*24*time.Hour))
+	if err != nil {
+		t.Fatal(err)
+	}
+	today := fixedTime().UTC().Format("2006-01-02")
+	oldDay := old.UTC().Format("2006-01-02")
+	if counts[today] != 2 {
+		t.Errorf("today count = %d, want 2 (got %v)", counts[today], counts)
+	}
+	if counts[oldDay] != 1 {
+		t.Errorf("old day count = %d, want 1 (got %v)", counts[oldDay], counts)
+	}
+}
+
+func TestAdminTaskStatsAndStorage(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	repo := NewAdminReadRepo(db)
+
+	// One completed task with a known duration, one failed.
+	job := seedJob(t, db, 2)
+	start := fixedTime().Add(-2 * time.Minute)
+	done := fixedTime().Add(-90 * time.Second)
+	queries := []string{
+		"UPDATE tasks SET status='completed', result_artifact_id=?, started_at=?, completed_at=? WHERE job_id=? AND chunk_index=0",
+		"UPDATE tasks SET status='failed' WHERE job_id=? AND chunk_index=1",
+	}
+	for i, q := range queries {
+		args := []any{uuid.NewString(), start.UnixNano(), done.UnixNano(), job.ID.String()}
+		if i == 1 {
+			args = []any{job.ID.String()}
+		}
+		if _, err := db.ExecContext(ctx, q, args...); err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	completed, failed, avg, err := repo.TaskStats(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if completed != 1 || failed != 1 {
+		t.Errorf("stats = completed %d failed %d, want 1/1", completed, failed)
+	}
+	if avg < 29 || avg > 31 {
+		t.Errorf("avg duration = %.1fs, want ~30s", avg)
+	}
+
+	// Artifact sizes by kind.
+	for _, kind := range []string{"input", "shard", "final_result"} {
+		if _, err := db.ExecContext(ctx, "INSERT INTO artifacts (id, job_id, kind, filename, storage_key, content_type, size_bytes, created_at) VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
+			uuid.NewString(), job.ID.String(), kind, kind+".csv", "key-"+kind, "text/csv", int64(len(kind)*1000), fixedTime().UnixNano()); err != nil {
+			t.Fatal(err)
+		}
+	}
+	sizes, err := repo.ArtifactSizeByKind(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if sizes["input"] != 5000 || sizes["shard"] != 5000 || sizes["final_result"] != 12000 {
+		t.Errorf("sizes = %v", sizes)
+	}
+	dbBytes, err := repo.DatabaseSizeBytes(ctx)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if dbBytes <= 0 {
+		t.Errorf("database size = %d, want > 0", dbBytes)
+	}
+}
@@ -0,0 +1,91 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"errors"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+// ArtifactRepo implements usecase.ArtifactRepository on SQLite.
+type ArtifactRepo struct {
+	db *sql.DB
+}
+
+func NewArtifactRepo(db *sql.DB) *ArtifactRepo {
+	return &ArtifactRepo{db: db}
+}
+
+const artifactColumns = `id, job_id, task_id, attempt, kind, filename, storage_key,
+	content_type, size_bytes, sha256, created_at`
+
+// scanArtifact maps one row onto a domain.Artifact.
+func scanArtifact(row interface{ Scan(dest ...any) error }) (*domain.Artifact, error) {
+	var (
+		a    domain.Artifact
+		kind string
+	)
+	var (
+		taskID    sql.NullString
+		attempt   sql.NullInt64
+		createdAt sql.NullInt64
+	)
+	if err := row.Scan(
+		&a.ID, &a.JobID, &taskID, &attempt, &kind, &a.Filename, &a.StorageKey,
+		&a.ContentType, &a.SizeBytes, &a.SHA256, &createdAt,
+	); err != nil {
+		return nil, err
+	}
+	a.CreatedAt = decodeTime(createdAt.Int64)
+	a.Kind = domain.ArtifactKind(kind)
+	if taskID.Valid {
+		if id, err := uuid.Parse(taskID.String); err == nil {
+			a.TaskID = &id
+		}
+	}
+	if attempt.Valid {
+		value := int(attempt.Int64)
+		a.Attempt = &value
+	}
+	return &a, nil
+}
+
+func (r *ArtifactRepo) Insert(ctx context.Context, a *domain.Artifact) error {
+	_, err := conn(ctx, r.db).ExecContext(ctx, `
+INSERT INTO artifacts (id, job_id, task_id, attempt, kind, filename, storage_key,
+	content_type, size_bytes, sha256, created_at)
+VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
+		a.ID.String(), a.JobID.String(), nullableUUID(a.TaskID), nullableInt(a.Attempt),
+		string(a.Kind), a.Filename, a.StorageKey, a.ContentType, a.SizeBytes, a.SHA256,
+		encodeTime(a.CreatedAt))
+	return err
+}
+
+func (r *ArtifactRepo) Get(ctx context.Context, id uuid.UUID) (*domain.Artifact, error) {
+	row := conn(ctx, r.db).QueryRowContext(ctx,
+		"SELECT "+artifactColumns+" FROM artifacts WHERE id = ?", id.String())
+	artifact, err := scanArtifact(row)
+	return artifact, mapErrNoRows(err, domain.ErrArtifactNotFound)
+}
+
+func (r *ArtifactRepo) FindPartialResult(ctx context.Context, taskID uuid.UUID, attempt int) (*domain.Artifact, error) {
+	row := conn(ctx, r.db).QueryRowContext(ctx,
+		"SELECT "+artifactColumns+" FROM artifacts WHERE task_id = ? AND attempt = ? AND kind = ?",
+		taskID.String(), attempt, string(domain.ArtifactPartialResult))
+	artifact, err := scanArtifact(row)
+	if errors.Is(err, sql.ErrNoRows) {
+		return nil, nil
+	}
+	return artifact, err
+}
+
+// nullableInt renders a nilable int as its value, or NULL.
+func nullableInt(n *int) any {
+	if n == nil {
+		return nil
+	}
+	return *n
+}
@@ -0,0 +1,196 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+// JobRepo implements usecase.JobRepository on SQLite.
+type JobRepo struct {
+	db *sql.DB
+}
+
+func NewJobRepo(db *sql.DB) *JobRepo {
+	return &JobRepo{db: db}
+}
+
+const jobColumns = `id, workload, input_uri, parameters, status, created_at, completed_at,
+	input_artifact_id, result_artifact_id, error_code, error_message, reducer_started_at, owner_id`
+
+// scanJob maps one row onto a domain.Job. Scanned values follow the sqlite
+// column order exactly: ids are TEXT, parameters JSON TEXT, timestamps unix
+// nanoseconds (nullable), statuses plain strings.
+func scanJob(row interface{ Scan(dest ...any) error }) (*domain.Job, error) {
+	var (
+		j      domain.Job
+		status string
+		params string
+	)
+	var (
+		createdAt                              sql.NullInt64
+		completedAt, reducerStartedAt          sql.NullInt64
+		inputArtifact, resultArtifact, ownerID sql.NullString
+		errorCode, errorMessage                sql.NullString
+	)
+	if err := row.Scan(
+		&j.ID, &j.Workload, &j.InputURI, &params, &status, &createdAt,
+		&completedAt, &inputArtifact, &resultArtifact, &errorCode, &errorMessage,
+		&reducerStartedAt, &ownerID,
+	); err != nil {
+		return nil, err
+	}
+	if err := decodeJSON(params, &j.Parameters); err != nil {
+		return nil, err
+	}
+	j.Status = domain.JobStatus(status)
+	j.CreatedAt = decodeTime(createdAt.Int64)
+	if completedAt.Valid {
+		value := decodeTime(completedAt.Int64)
+		j.CompletedAt = &value
+	}
+	if reducerStartedAt.Valid {
+		value := decodeTime(reducerStartedAt.Int64)
+		j.ReducerStartedAt = &value
+	}
+	if inputArtifact.Valid {
+		if id, err := uuid.Parse(inputArtifact.String); err == nil {
+			j.InputArtifactID = &id
+		}
+	}
+	if resultArtifact.Valid {
+		if id, err := uuid.Parse(resultArtifact.String); err == nil {
+			j.ResultArtifactID = &id
+		}
+	}
+	if ownerID.Valid {
+		if id, err := uuid.Parse(ownerID.String); err == nil {
+			j.OwnerID = &id
+		}
+	}
+	if errorCode.Valid {
+		j.ErrorCode = &errorCode.String
+	}
+	if errorMessage.Valid {
+		j.ErrorMessage = &errorMessage.String
+	}
+	return &j, nil
+}
+
+// Insert runs inside the caller's transaction alongside the job's tasks.
+func (r *JobRepo) Insert(ctx context.Context, j *domain.Job) error {
+	_, err := conn(ctx, r.db).ExecContext(ctx, `
+INSERT INTO jobs (id, workload, input_uri, parameters, status, created_at, owner_id)
+VALUES (?, ?, ?, ?, ?, ?, ?)`,
+		j.ID.String(), j.Workload, j.InputURI, encodeJSON(j.Parameters), string(j.Status),
+		encodeTime(j.CreatedAt), nullableUUID(j.OwnerID))
+	return err
+}
+
+func (r *JobRepo) Get(ctx context.Context, id uuid.UUID) (*domain.Job, error) {
+	row := conn(ctx, r.db).QueryRowContext(ctx,
+		"SELECT "+jobColumns+" FROM jobs WHERE id = ?", id.String())
+	job, err := scanJob(row)
+	return job, mapErrNoRows(err, domain.ErrJobNotFound)
+}
+
+func (r *JobRepo) ClaimReduction(ctx context.Context, id uuid.UUID, startedAt time.Time) (bool, error) {
+	res, err := conn(ctx, r.db).ExecContext(ctx, `
+UPDATE jobs SET reducer_started_at = ?
+WHERE id = ? AND status = ? AND reducer_started_at IS NULL`,
+		encodeTime(startedAt), id.String(), string(domain.JobReducing))
+	if err != nil {
+		return false, err
+	}
+	affected, err := res.RowsAffected()
+	return affected == 1, err
+}
+
+func (r *JobRepo) CompleteWithResult(ctx context.Context, id, resultArtifactID uuid.UUID, completedAt time.Time) error {
+	res, err := conn(ctx, r.db).ExecContext(ctx, `
+UPDATE jobs SET status = ?, result_artifact_id = ?, completed_at = ?,
+	reducer_started_at = NULL, error_code = NULL, error_message = NULL
+WHERE id = ? AND status = ?`,
+		string(domain.JobCompleted), resultArtifactID.String(), encodeTime(completedAt),
+		id.String(), string(domain.JobReducing))
+	if err != nil {
+		return err
+	}
+	affected, err := res.RowsAffected()
+	if err != nil {
+		return err
+	}
+	if affected == 0 {
+		return domain.ErrJobNotFound
+	}
+	return nil
+}
+
+func (r *JobRepo) FailReduction(ctx context.Context, id uuid.UUID, code, message string, completedAt time.Time) error {
+	_, err := conn(ctx, r.db).ExecContext(ctx, `
+UPDATE jobs SET status = ?, completed_at = ?, error_code = ?, error_message = ?,
+	reducer_started_at = NULL
+WHERE id = ? AND status = ?`,
+		string(domain.JobFailed), encodeTime(completedAt), code, message,
+		id.String(), string(domain.JobReducing))
+	return err
+}
+
+func (r *JobRepo) UpdateStatus(ctx context.Context, id uuid.UUID,
+	status domain.JobStatus, completedAt *time.Time) error {
+
+	res, err := conn(ctx, r.db).ExecContext(ctx,
+		"UPDATE jobs SET status = ?, completed_at = ? WHERE id = ?",
+		string(status), encodeTimePtr(completedAt), id.String())
+	if err != nil {
+		return err
+	}
+	affected, err := res.RowsAffected()
+	if err != nil {
+		return err
+	}
+	if affected == 0 {
+		return domain.ErrJobNotFound
+	}
+	return nil
+}
+
+// nullableUUID renders a nilable UUID as its text, or NULL.
+func nullableUUID(id *uuid.UUID) any {
+	if id == nil {
+		return nil
+	}
+	return id.String()
+}
+
+// ListCompletedBefore returns jobs whose completion timestamp is older than
+// the cutoff (completed and failed both count as finished).
+func (r *JobRepo) ListCompletedBefore(ctx context.Context, cutoff time.Time) ([]domain.Job, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+jobColumns+" FROM jobs WHERE completed_at IS NOT NULL AND completed_at < ? ORDER BY completed_at ASC",
+		encodeTime(cutoff))
+	if err != nil {
+		return nil, fmt.Errorf("list completed jobs: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	var jobs []domain.Job
+	for rows.Next() {
+		job, err := scanJob(rows)
+		if err != nil {
+			return nil, err
+		}
+		jobs = append(jobs, *job)
+	}
+	return jobs, rows.Err()
+}
+
+// Delete removes the job row; tasks, artifacts and task_results cascade.
+func (r *JobRepo) Delete(ctx context.Context, id uuid.UUID) error {
+	_, err := conn(ctx, r.db).ExecContext(ctx, "DELETE FROM jobs WHERE id = ?", id.String())
+	return err
+}
@@ -0,0 +1,85 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"embed"
+	"fmt"
+	"log/slog"
+	"regexp"
+	"sort"
+	"strconv"
+)
+
+//go:embed migrations/*.sql
+var migrationFiles embed.FS
+
+var migrationNamePattern = regexp.MustCompile(`^([0-9]+)_[a-z0-9_]+\.sql$`)
+
+// Migrate applies every embedded migration above the current PRAGMA
+// user_version watermark, each inside its own transaction. It is idempotent:
+// the watermark only advances after a migration commits.
+func Migrate(ctx context.Context, db *sql.DB, log *slog.Logger) error {
+	entries, err := migrationFiles.ReadDir("migrations")
+	if err != nil {
+		return fmt.Errorf("read embedded migrations: %w", err)
+	}
+	type file struct {
+		version int
+		name    string
+	}
+	var files []file
+	byVersion := map[int]string{}
+	for _, entry := range entries {
+		match := migrationNamePattern.FindStringSubmatch(entry.Name())
+		if match == nil {
+			continue
+		}
+		version, err := strconv.Atoi(match[1])
+		if err != nil {
+			return fmt.Errorf("migration %q has an invalid version: %w", entry.Name(), err)
+		}
+		if _, duplicate := byVersion[version]; duplicate {
+			return fmt.Errorf("migration version %d is duplicated", version)
+		}
+		body, err := migrationFiles.ReadFile("migrations/" + entry.Name())
+		if err != nil {
+			return fmt.Errorf("read migration %q: %w", entry.Name(), err)
+		}
+		byVersion[version] = string(body)
+		files = append(files, file{version: version, name: entry.Name()})
+	}
+	if len(files) == 0 {
+		return fmt.Errorf("no sqlite migrations are embedded")
+	}
+	sort.Slice(files, func(i, j int) bool { return files[i].version < files[j].version })
+
+	var applied int
+	if err := db.QueryRowContext(ctx, "PRAGMA user_version").Scan(&applied); err != nil {
+		return fmt.Errorf("read schema version: %w", err)
+	}
+	for _, item := range files {
+		if item.version <= applied {
+			continue
+		}
+		if log != nil {
+			log.Info("applying sqlite migration", "version", item.version, "file", item.name)
+		}
+		tx, err := db.BeginTx(ctx, nil)
+		if err != nil {
+			return err
+		}
+		if _, err := tx.ExecContext(ctx, byVersion[item.version]); err != nil {
+			_ = tx.Rollback()
+			return fmt.Errorf("apply migration %s: %w", item.name, err)
+		}
+		if _, err := tx.ExecContext(ctx, fmt.Sprintf("PRAGMA user_version = %d", item.version)); err != nil {
+			_ = tx.Rollback()
+			return fmt.Errorf("advance schema version after %s: %w", item.name, err)
+		}
+		if err := tx.Commit(); err != nil {
+			return fmt.Errorf("commit migration %s: %w", item.name, err)
+		}
+	}
+	return nil
+}
@@ -0,0 +1,96 @@
+-- 0001: core schema. SQLite stores enums as TEXT with CHECK constraints and
+-- JSON documents as TEXT; timestamps are unix nanoseconds (INTEGER).
+CREATE TABLE IF NOT EXISTS jobs (
+    id                  TEXT PRIMARY KEY,
+    workload            TEXT NOT NULL,
+    input_uri           TEXT NOT NULL DEFAULT '',
+    parameters          TEXT NOT NULL DEFAULT '{}',
+    status              TEXT NOT NULL DEFAULT 'pending'
+                        CHECK (status IN ('pending','running','reducing','completed','failed','cancelled')),
+    created_at          INTEGER NOT NULL,
+    completed_at        INTEGER,
+    input_artifact_id   TEXT,
+    result_artifact_id  TEXT,
+    error_code          TEXT,
+    error_message       TEXT,
+    reducer_started_at  INTEGER,
+    owner_id            TEXT
+);
+
+CREATE TABLE IF NOT EXISTS tasks (
+    id                  TEXT PRIMARY KEY,
+    job_id              TEXT NOT NULL REFERENCES jobs(id) ON DELETE CASCADE,
+    chunk_index         INTEGER NOT NULL,
+    workload            TEXT NOT NULL,
+    input_uri           TEXT,
+    input_artifact_id   TEXT,
+    input_sha256        TEXT NOT NULL,
+    parameters          TEXT NOT NULL DEFAULT '{}',
+    status              TEXT NOT NULL DEFAULT 'pending'
+                        CHECK (status IN ('pending','leased','running','completed','failed','cancelled')),
+    attempt             INTEGER NOT NULL DEFAULT 0 CHECK (attempt >= 0),
+    max_attempts        INTEGER NOT NULL DEFAULT 3 CHECK (max_attempts > 0),
+    lease_owner         TEXT,
+    lease_expires_at    INTEGER,
+    result_artifact_id  TEXT,
+    metrics             TEXT,
+    error_code          TEXT,
+    error_message       TEXT,
+    created_at          INTEGER NOT NULL,
+    started_at          INTEGER,
+    completed_at        INTEGER,
+    version             INTEGER NOT NULL DEFAULT 0,
+    CONSTRAINT uq_tasks_job_chunk UNIQUE (job_id, chunk_index),
+    CONSTRAINT ck_tasks_completed_result CHECK (
+        status <> 'completed' OR (result_artifact_id IS NOT NULL)
+    ),
+    CONSTRAINT ck_tasks_leased_owner CHECK (
+        status <> 'leased' OR (lease_owner IS NOT NULL AND lease_expires_at IS NOT NULL)
+    )
+);
+
+CREATE INDEX IF NOT EXISTS ix_tasks_claim ON tasks (status, lease_expires_at, created_at);
+CREATE INDEX IF NOT EXISTS ix_tasks_job ON tasks (job_id);
+
+CREATE TABLE IF NOT EXISTS artifacts (
+    id            TEXT PRIMARY KEY,
+    job_id        TEXT NOT NULL REFERENCES jobs(id) ON DELETE CASCADE,
+    task_id       TEXT,
+    attempt       INTEGER,
+    kind          TEXT NOT NULL
+                  CHECK (kind IN ('input','shard','partial_result','final_result','log')),
+    filename      TEXT NOT NULL,
+    storage_key   TEXT NOT NULL,
+    content_type  TEXT NOT NULL,
+    size_bytes    INTEGER NOT NULL DEFAULT 0,
+    sha256        TEXT NOT NULL DEFAULT '',
+    created_at    INTEGER NOT NULL,
+    CONSTRAINT uq_partial_result_task_attempt UNIQUE (task_id, attempt)
+);
+
+CREATE INDEX IF NOT EXISTS ix_artifacts_job ON artifacts (job_id);
+
+CREATE TABLE IF NOT EXISTS workers (
+    id                 TEXT PRIMARY KEY,
+    name               TEXT NOT NULL,
+    capabilities       TEXT NOT NULL DEFAULT '[]',
+    status             TEXT NOT NULL DEFAULT 'online'
+                       CHECK (status IN ('online','busy','offline')),
+    owner_id           TEXT,
+    trust_level        TEXT NOT NULL DEFAULT 'trusted'
+                       CHECK (trust_level IN ('trusted','untrusted')),
+    last_heartbeat_at  INTEGER NOT NULL,
+    created_at         INTEGER NOT NULL,
+    updated_at         INTEGER NOT NULL
+);
+
+CREATE TABLE IF NOT EXISTS task_results (
+    task_id             TEXT NOT NULL REFERENCES tasks(id) ON DELETE CASCADE,
+    owner_id            TEXT NOT NULL,
+    result_sha256       TEXT NOT NULL,
+    result_artifact_id  TEXT NOT NULL,
+    created_at          INTEGER NOT NULL DEFAULT 0,
+    PRIMARY KEY (task_id, owner_id)
+);
+
+CREATE INDEX IF NOT EXISTS ix_task_results_task ON task_results (task_id);
@@ -0,0 +1,8 @@
+-- 0002: per-workload enable/disable. Absence of a row means "enabled" (the
+-- catalog default); a row only exists once an admin flipped a workload off or
+-- back on.
+CREATE TABLE IF NOT EXISTS workload_settings (
+    workload   TEXT NOT NULL PRIMARY KEY,
+    enabled    INTEGER NOT NULL CHECK (enabled IN (0, 1)),
+    updated_at INTEGER NOT NULL
+);
@@ -0,0 +1,172 @@
+// Package sqlite implements the usecase repository ports on an embedded
+// SQLite database. It is the single-binary storage backend: no external
+// service, one file per database, pure-Go driver (modernc.org/sqlite) so the
+// static release binaries stay static.
+//
+// Concurrency model: SQLite allows exactly one writer. Every repository write
+// runs inside a TxManager transaction, and the database is opened with a
+// busy_timeout, so concurrent writers serialize instead of failing. The
+// postgres claim path uses FOR UPDATE SKIP LOCKED; here the same guarantee
+// comes from the write lock of the surrounding transaction — ClaimNext is
+// always called inside WithinTx by the usecase layer, so SELECT + UPDATE
+// cannot interleave.
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"time"
+
+	_ "modernc.org/sqlite"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// Open opens (and creates when missing) the database file, applying WAL,
+// foreign keys, and a busy timeout. Callers own the returned handle.
+func Open(path string) (*sql.DB, error) {
+	dsn := fmt.Sprintf("file:%s?_pragma=busy_timeout(10000)&_pragma=journal_mode(WAL)&_pragma=foreign_keys(1)&_pragma=synchronous(NORMAL)", path)
+	db, err := sql.Open("sqlite", dsn)
+	if err != nil {
+		return nil, fmt.Errorf("open sqlite database: %w", err)
+	}
+	if err := db.PingContext(context.Background()); err != nil {
+		_ = db.Close()
+		return nil, fmt.Errorf("ping sqlite database: %w", err)
+	}
+	return db, nil
+}
+
+// querier is satisfied by both *sql.DB and *sql.Tx, letting every repository
+// method run identically inside or outside a transaction.
+type querier interface {
+	ExecContext(ctx context.Context, query string, args ...any) (sql.Result, error)
+	QueryContext(ctx context.Context, query string, args ...any) (*sql.Rows, error)
+	QueryRowContext(ctx context.Context, query string, args ...any) *sql.Row
+}
+
+// txKey is an unexported struct type, so no other package can collide with it
+// or reach the transaction we stash in the context.
+type txKey struct{}
+
+// TxManager implements usecase.TxManager.
+type TxManager struct {
+	db *sql.DB
+}
+
+func NewTxManager(db *sql.DB) *TxManager {
+	return &TxManager{db: db}
+}
+
+var _ usecase.TxManager = (*TxManager)(nil)
+
+// WithinTx runs fn inside one transaction, committing on success and rolling
+// back on any error or panic. The transaction travels in the context, the
+// same pattern as the postgres backend. SQLite write transactions are
+// serialized by the database's single-writer lock, so a concurrent writer
+// waits on the busy timeout instead of racing.
+func (m *TxManager) WithinTx(ctx context.Context, fn func(ctx context.Context) error) error {
+	if _, ok := ctx.Value(txKey{}).(*sql.Tx); ok {
+		return fn(ctx)
+	}
+	tx, err := m.db.BeginTx(ctx, nil)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = tx.Rollback() }()
+
+	if err := fn(context.WithValue(ctx, txKey{}, tx)); err != nil {
+		return err
+	}
+	return tx.Commit()
+}
+
+// conn returns the transaction bound to ctx, or the database when there is none.
+func conn(ctx context.Context, db *sql.DB) querier {
+	if tx, ok := ctx.Value(txKey{}).(*sql.Tx); ok {
+		return tx
+	}
+	return db
+}
+
+// --- JSON and null helpers ------------------------------------------------
+
+// encodeJSON stores a Go value as JSON text, defaulting to "{}".
+func encodeJSON(value any) string {
+	if value == nil {
+		return "{}"
+	}
+	encoded, err := json.Marshal(value)
+	if err != nil {
+		return "{}"
+	}
+	return string(encoded)
+}
+
+// decodeJSON reads a JSON text column into the destination.
+func decodeJSON(raw any, destination any) error {
+	text, ok := raw.(string)
+	if !ok {
+		return nil
+	}
+	if text == "" {
+		return nil
+	}
+	return json.Unmarshal([]byte(text), destination)
+}
+
+// encodeTime stores a time as unix nanoseconds (NULL for zero time).
+func encodeTime(t time.Time) any {
+	if t.IsZero() {
+		return nil
+	}
+	return t.UnixNano()
+}
+
+// encodeTimePtr stores a nilable time as unix nanoseconds.
+func encodeTimePtr(t *time.Time) any {
+	if t == nil {
+		return nil
+	}
+	return t.UnixNano()
+}
+
+// decodeTime reads a unix-nanosecond column back into a time.Time.
+func decodeTime(raw any) time.Time {
+	switch v := raw.(type) {
+	case int64:
+		return time.Unix(0, v).UTC()
+	case int:
+		return time.Unix(0, int64(v)).UTC()
+	}
+	return time.Time{}
+}
+
+// nullIfEmpty maps "" to SQL NULL.
+func nullIfEmpty(s string) any {
+	if s == "" {
+		return nil
+	}
+	return s
+}
+
+// mapErrNoRows translates sql.ErrNoRows into the domain not-found errors.
+func mapErrNoRows(err error, notFound error) error {
+	if errors.Is(err, sql.ErrNoRows) {
+		return notFound
+	}
+	return err
+}
+
+var (
+	_ usecase.JobRepository        = (*JobRepo)(nil)
+	_ usecase.TaskRepository       = (*TaskRepo)(nil)
+	_ usecase.ArtifactRepository   = (*ArtifactRepo)(nil)
+	_ usecase.WorkerRepository     = (*WorkerRepo)(nil)
+	_ usecase.TaskResultRepository = (*TaskResultRepo)(nil)
+	_ usecase.UIReadRepository     = (*UIReadRepo)(nil)
+	_ usecase.TxManager            = (*TxManager)(nil)
+)
@@ -0,0 +1,349 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"errors"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// newTestDB opens an isolated on-disk database and applies the migrations.
+func newTestDB(t *testing.T) *sql.DB {
+	t.Helper()
+	db, err := Open(filepath.Join(t.TempDir(), "test.db"))
+	if err != nil {
+		t.Fatalf("open: %v", err)
+	}
+	t.Cleanup(func() { _ = db.Close() })
+	if err := Migrate(context.Background(), db, nil); err != nil {
+		t.Fatalf("migrate: %v", err)
+	}
+	return db
+}
+
+func fixedTime() time.Time {
+	return time.Date(2026, 8, 2, 12, 0, 0, 0, time.UTC)
+}
+
+func seedJob(t *testing.T, db *sql.DB, n int) *domain.Job {
+	t.Helper()
+	ctx := context.Background()
+	tx := NewTxManager(db)
+	chunks := make([]domain.ChunkSpec, 0, n)
+	for i := 0; i < n; i++ {
+		chunks = append(chunks, domain.ChunkSpec{
+			ChunkIndex:  i,
+			InputURI:    "s3://chunk-" + string(rune('a'+i)),
+			InputSHA256: "sha-" + string(rune('a'+i)),
+		})
+	}
+	job, tasks, err := domain.NewJobWithTasks("similarity_search", "s3://ds", nil, chunks, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := tx.WithinTx(ctx, func(ctx context.Context) error {
+		if err := NewJobRepo(db).Insert(ctx, job); err != nil {
+			return err
+		}
+		return NewTaskRepo(db).InsertBatch(ctx, tasks)
+	}); err != nil {
+		t.Fatalf("seed: %v", err)
+	}
+	return job
+}
+
+func TestMigrateIsIdempotent(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	if err := Migrate(ctx, db, nil); err != nil {
+		t.Fatalf("second migrate: %v", err)
+	}
+	var version int
+	if err := db.QueryRowContext(ctx, "PRAGMA user_version").Scan(&version); err != nil {
+		t.Fatal(err)
+	}
+	if version != 2 {
+		t.Errorf("user_version = %d, want 2", version)
+	}
+}
+
+func TestJobRepoRoundTrip(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 1)
+
+	got, err := NewJobRepo(db).Get(ctx, job.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.Workload != job.Workload || got.Status != domain.JobPending {
+		t.Errorf("job = %+v", got)
+	}
+	if err := NewJobRepo(db).UpdateStatus(ctx, job.ID, domain.JobRunning, nil); err != nil {
+		t.Fatal(err)
+	}
+	got, err = NewJobRepo(db).Get(ctx, job.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.Status != domain.JobRunning {
+		t.Errorf("status = %q, want running", got.Status)
+	}
+	if _, err := NewJobRepo(db).Get(ctx, uuid.New()); !errors.Is(err, domain.ErrJobNotFound) {
+		t.Errorf("missing job err = %v, want ErrJobNotFound", err)
+	}
+}
+
+func TestClaimGivesEachTaskToExactlyOneWorker(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	seedJob(t, db, 3)
+	repo := NewTaskRepo(db)
+
+	claimed := map[uuid.UUID]bool{}
+	for i := 0; i < 3; i++ {
+		task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{
+			Workloads:  []string{"similarity_search"},
+			Owner:      "w1",
+			Now:        fixedTime(),
+			LeaseUntil: fixedTime().Add(time.Minute),
+		})
+		if err != nil {
+			t.Fatal(err)
+		}
+		if task == nil {
+			t.Fatal("claim returned nil on a non-empty queue")
+		}
+		if claimed[task.ID] {
+			t.Fatalf("task %s claimed twice", task.ID)
+		}
+		claimed[task.ID] = true
+		if task.Status != domain.TaskLeased || task.Attempt != 1 || task.LeaseOwner == nil || *task.LeaseOwner != "w1" {
+			t.Errorf("task = %+v", task)
+		}
+	}
+	task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: []string{"similarity_search"}, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(time.Minute)})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if task != nil {
+		t.Fatal("claim must return nil on an empty queue")
+	}
+}
+
+func TestUpdateRejectsStaleVersion(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	seedJob(t, db, 1)
+	repo := NewTaskRepo(db)
+	task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(time.Minute)})
+	if err != nil || task == nil {
+		t.Fatalf("claim: %v", err)
+	}
+	stale := *task
+	task.Status = domain.TaskRunning
+	task.Version++ // as a domain method would have done
+	if err := repo.Update(ctx, task); err != nil {
+		t.Fatal(err)
+	}
+	stale.Status = domain.TaskCompleted
+	stale.Version++
+	if err := repo.Update(ctx, &stale); !errors.Is(err, domain.ErrLeaseConflict) {
+		t.Errorf("stale update err = %v, want ErrLeaseConflict", err)
+	}
+}
+
+func TestExpireLeasesRequeuesElapsedTasks(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	seedJob(t, db, 1)
+	repo := NewTaskRepo(db)
+	task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(-time.Minute)})
+	if err != nil || task == nil {
+		t.Fatalf("claim: %v", err)
+	}
+	affected, err := repo.ExpireLeases(ctx, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(affected) != 1 {
+		t.Fatalf("affected = %v, want 1 job", affected)
+	}
+	task, err = repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w2", Now: fixedTime(), LeaseUntil: fixedTime().Add(time.Minute)})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if task == nil || task.Attempt != 2 {
+		t.Errorf("requeued task = %+v, want attempt 2", task)
+	}
+}
+
+func TestExpireLeasesFailsAfterFinalAttempt(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	seedJob(t, db, 1)
+	repo := NewTaskRepo(db)
+	for attempt := 1; attempt <= 3; attempt++ {
+		task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(-time.Minute)})
+		if err != nil || task == nil {
+			t.Fatalf("claim %d: %v", attempt, err)
+		}
+		if _, err := repo.ExpireLeases(ctx, fixedTime()); err != nil {
+			t.Fatal(err)
+		}
+	}
+	task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(time.Minute)})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if task != nil {
+		t.Fatal("exhausted task must not be claimable")
+	}
+}
+
+func TestArtifactRepoRoundTripAndUniqueness(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 1)
+	repo := NewArtifactRepo(db)
+	attempt := 1
+	artifact, err := domain.NewArtifact(job.ID, nil, domain.ArtifactShard, "shard-0.tsv", "text/tab-separated-values", fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	artifact.SetContent("abc123", 42)
+	if err := repo.Insert(ctx, artifact); err != nil {
+		t.Fatal(err)
+	}
+	got, err := repo.Get(ctx, artifact.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.SHA256 != "abc123" || got.SizeBytes != 42 {
+		t.Errorf("artifact = %+v", got)
+	}
+
+	partialTask := uuid.New()
+	partial, err := domain.NewArtifact(job.ID, &partialTask, domain.ArtifactPartialResult, "p.csv", "text/csv", fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	partial.Attempt = &attempt
+	if err := repo.Insert(ctx, partial); err != nil {
+		t.Fatal(err)
+	}
+	found, err := repo.FindPartialResult(ctx, partialTask, attempt)
+	if err != nil || found == nil {
+		t.Fatalf("find partial: %v", err)
+	}
+	duplicate, err := domain.NewArtifact(job.ID, &partialTask, domain.ArtifactPartialResult, "p2.csv", "text/csv", fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	duplicate.Attempt = &attempt
+	if err := repo.Insert(ctx, duplicate); err == nil {
+		t.Fatal("duplicate partial for the same attempt must fail")
+	}
+}
+
+func TestWorkerRepoRoundTripAndLiveness(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	repo := NewWorkerRepo(db)
+	worker, err := domain.NewWorker("w1", []string{"similarity-search"}, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.Insert(ctx, worker); err != nil {
+		t.Fatal(err)
+	}
+	got, err := repo.Get(ctx, worker.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.Name != "w1" || len(got.Capabilities) != 1 || got.TrustLevel != domain.WorkerTrusted {
+		t.Errorf("worker = %+v", got)
+	}
+	if err := repo.Touch(ctx, worker.ID, fixedTime().Add(time.Hour)); err != nil {
+		t.Fatal(err)
+	}
+	changed, err := repo.MarkStaleOffline(ctx, fixedTime().Add(2*time.Hour))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if changed != 1 {
+		t.Errorf("offline changes = %d, want 1", changed)
+	}
+	got, err = repo.Get(ctx, worker.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.Status != domain.WorkerOffline {
+		t.Errorf("status = %q, want offline", got.Status)
+	}
+}
+
+func TestTaskResultVotes(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 1)
+	repo := NewTaskResultRepo(db)
+	task, err := domain.NewTask(job.ID, 1, "similarity_search", "s3://in", "sha", nil, 3, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := NewTaskRepo(db).InsertBatch(ctx, []*domain.Task{task}); err != nil {
+		t.Fatal(err)
+	}
+	taskID := task.ID
+	artifactID := uuid.New()
+	ownerA, ownerB := uuid.New(), uuid.New()
+	if err := repo.RecordVote(ctx, taskID, ownerA, "hash", artifactID); err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.RecordVote(ctx, taskID, ownerB, "hash", artifactID); err != nil {
+		t.Fatal(err)
+	}
+	if err := repo.RecordVote(ctx, taskID, ownerA, "hash2", artifactID); err != nil {
+		t.Fatal(err)
+	}
+	n, err := repo.CountAgreeing(ctx, taskID, "hash")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if n != 1 {
+		t.Errorf("agreeing = %d, want 1 (owner A changed its vote)", n)
+	}
+}
+
+func TestCancelByJobInvalidatesTasks(t *testing.T) {
+	db := newTestDB(t)
+	ctx := context.Background()
+	job := seedJob(t, db, 2)
+	repo := NewTaskRepo(db)
+	task, err := repo.ClaimNext(ctx, usecase.ClaimFilter{Workloads: nil, Owner: "w1", Now: fixedTime(), LeaseUntil: fixedTime().Add(time.Minute)})
+	if err != nil || task == nil {
+		t.Fatalf("claim: %v", err)
+	}
+	cancelled, err := repo.CancelByJob(ctx, job.ID, fixedTime())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if cancelled != 2 {
+		t.Errorf("cancelled = %d, want 2", cancelled)
+	}
+	got, err := repo.Get(ctx, task.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got.Status != domain.TaskCancelled || got.LeaseOwner != nil {
+		t.Errorf("cancelled task = %+v", got)
+	}
+}
@@ -0,0 +1,63 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+// Known statuses per entity, so counts are zero-filled and every status is
+// always present in the metrics (a flat 0 line beats a gap on the dashboard).
+var (
+	taskStatuses   = []string{string(domain.TaskPending), string(domain.TaskLeased), string(domain.TaskRunning), string(domain.TaskCompleted), string(domain.TaskFailed), string(domain.TaskCancelled)}
+	jobStatuses    = []string{string(domain.JobPending), string(domain.JobRunning), string(domain.JobReducing), string(domain.JobCompleted), string(domain.JobFailed), string(domain.JobCancelled)}
+	workerStatuses = []string{string(domain.WorkerOnline), string(domain.WorkerBusy), string(domain.WorkerOffline)}
+)
+
+// StatsRepo answers the aggregate status counts the business metrics report.
+type StatsRepo struct {
+	db *sql.DB
+}
+
+func NewStatsRepo(db *sql.DB) *StatsRepo {
+	return &StatsRepo{db: db}
+}
+
+// Counts returns status->count maps for tasks, jobs, and workers, each
+// zero-filled across its known statuses.
+func (r *StatsRepo) Counts(ctx context.Context) (tasks, jobs, workers map[string]int, err error) {
+	if tasks, err = r.countByStatus(ctx, "tasks", taskStatuses); err != nil {
+		return nil, nil, nil, err
+	}
+	if jobs, err = r.countByStatus(ctx, "jobs", jobStatuses); err != nil {
+		return nil, nil, nil, err
+	}
+	if workers, err = r.countByStatus(ctx, "workers", workerStatuses); err != nil {
+		return nil, nil, nil, err
+	}
+	return tasks, jobs, workers, nil
+}
+
+func (r *StatsRepo) countByStatus(ctx context.Context, table string, known []string) (map[string]int, error) {
+	out := make(map[string]int, len(known))
+	for _, s := range known {
+		out[s] = 0 // zero-fill
+	}
+	// table is a fixed internal constant, never user input — safe to format.
+	rows, err := r.db.QueryContext(ctx, fmt.Sprintf("SELECT status, count(*) FROM %s GROUP BY status", table))
+	if err != nil {
+		return nil, fmt.Errorf("count %s by status: %w", table, err)
+	}
+	defer func() { _ = rows.Close() }()
+	for rows.Next() {
+		var status string
+		var n int
+		if err := rows.Scan(&status, &n); err != nil {
+			return nil, err
+		}
+		out[status] = n
+	}
+	return out, rows.Err()
+}
@@ -0,0 +1,314 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"errors"
+	"strings"
+	"time"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+	"github.com/emil28092005/SciMesh/coordinator/internal/usecase"
+)
+
+// TaskRepo implements usecase.TaskRepository on SQLite.
+type TaskRepo struct {
+	db *sql.DB
+}
+
+func NewTaskRepo(db *sql.DB) *TaskRepo {
+	return &TaskRepo{db: db}
+}
+
+const taskColumns = `id, job_id, chunk_index, workload, input_uri, input_artifact_id, input_sha256,
+	parameters, status, attempt, max_attempts, lease_owner, lease_expires_at,
+	result_artifact_id, metrics, error_code, error_message,
+	created_at, started_at, completed_at, version`
+
+// scanTask maps one row onto a domain.Task.
+func scanTask(row interface{ Scan(dest ...any) error }) (*domain.Task, error) {
+	var (
+		t       domain.Task
+		status  string
+		params  string
+		metrics sql.NullString
+	)
+	var (
+		inputURI, leaseOwner, errorCode, errorMessage sql.NullString
+		inputArtifact, resultArtifact                 sql.NullString
+		leaseExpiresAt, startedAt, completedAt        sql.NullInt64
+		createdAt                                     sql.NullInt64
+	)
+	if err := row.Scan(
+		&t.ID, &t.JobID, &t.ChunkIndex, &t.Workload, &inputURI, &inputArtifact, &t.InputSHA256,
+		&params, &status, &t.Attempt, &t.MaxAttempts, &leaseOwner, &leaseExpiresAt,
+		&resultArtifact, &metrics, &errorCode, &errorMessage,
+		&createdAt, &startedAt, &completedAt, &t.Version,
+	); err != nil {
+		return nil, err
+	}
+	if err := decodeJSON(params, &t.Parameters); err != nil {
+		return nil, err
+	}
+	if metrics.Valid && metrics.String != "" {
+		if err := decodeJSON(metrics.String, &t.Metrics); err != nil {
+			return nil, err
+		}
+	}
+	t.Status = domain.TaskStatus(status)
+	t.CreatedAt = decodeTime(createdAt.Int64)
+	if inputURI.Valid {
+		t.InputURI = inputURI.String
+	}
+	if leaseOwner.Valid {
+		t.LeaseOwner = &leaseOwner.String
+	}
+	if inputArtifact.Valid {
+		if id, err := uuid.Parse(inputArtifact.String); err == nil {
+			t.InputArtifactID = &id
+		}
+	}
+	if resultArtifact.Valid {
+		if id, err := uuid.Parse(resultArtifact.String); err == nil {
+			t.ResultArtifactID = &id
+		}
+	}
+	if leaseExpiresAt.Valid {
+		value := decodeTime(leaseExpiresAt.Int64)
+		t.LeaseExpiresAt = &value
+	}
+	if startedAt.Valid {
+		value := decodeTime(startedAt.Int64)
+		t.StartedAt = &value
+	}
+	if completedAt.Valid {
+		value := decodeTime(completedAt.Int64)
+		t.CompletedAt = &value
+	}
+	if errorCode.Valid {
+		t.ErrorCode = &errorCode.String
+	}
+	if errorMessage.Valid {
+		t.ErrorMessage = &errorMessage.String
+	}
+	return &t, nil
+}
+
+// ClaimNext atomically leases the next eligible task. SQLite has no SKIP
+// LOCKED: the guarantee comes from the surrounding transaction's write lock —
+// the usecase layer always calls ClaimNext inside WithinTx, so SELECT + UPDATE
+// cannot interleave with another claimant.
+func (r *TaskRepo) ClaimNext(ctx context.Context, f usecase.ClaimFilter) (*domain.Task, error) {
+	workloadClause := ""
+	workloadArgs := []any{}
+	if len(f.Workloads) > 0 {
+		placeholders := make([]string, 0, len(f.Workloads))
+		for _, w := range f.Workloads {
+			placeholders = append(placeholders, "?")
+			workloadArgs = append(workloadArgs, w)
+		}
+		workloadClause = " AND workload IN (" + strings.Join(placeholders, ", ") + ")"
+	}
+	voterClause := ""
+	var voterArg any
+	if f.VoterOwner != nil {
+		voterClause = " AND NOT EXISTS (SELECT 1 FROM task_results tr WHERE tr.task_id = tasks.id AND tr.owner_id = ?)"
+		voterArg = f.VoterOwner.String()
+	}
+	args := append([]any{f.Owner, f.LeaseUntil.UnixNano(), f.Now.UnixNano()}, workloadArgs...)
+	if f.VoterOwner != nil {
+		args = append(args, voterArg)
+	}
+	query := `
+UPDATE tasks SET
+	status = 'leased',
+	attempt = attempt + 1,
+	lease_owner = ?,
+	lease_expires_at = ?,
+	started_at = COALESCE(started_at, ?),
+	version = version + 1
+WHERE id IN (
+	SELECT id FROM tasks
+	WHERE status = 'pending' AND attempt < max_attempts` + workloadClause + voterClause + `
+	ORDER BY created_at, chunk_index
+	LIMIT 1
+)
+RETURNING ` + taskColumns
+
+	row := conn(ctx, r.db).QueryRowContext(ctx, query, args...)
+	task, err := scanTask(row)
+	if errors.Is(err, sql.ErrNoRows) {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, err
+	}
+	return task, nil
+}
+
+func (r *TaskRepo) Get(ctx context.Context, id uuid.UUID) (*domain.Task, error) {
+	row := conn(ctx, r.db).QueryRowContext(ctx,
+		"SELECT "+taskColumns+" FROM tasks WHERE id = ?", id.String())
+	task, err := scanTask(row)
+	return task, mapErrNoRows(err, domain.ErrTaskNotFound)
+}
+
+// GetForUpdate reads a task. SQLite serializes writers inside a transaction,
+// so no row lock is needed: the surrounding write transaction already isolates
+// the read-modify-write sequence.
+func (r *TaskRepo) GetForUpdate(ctx context.Context, id uuid.UUID) (*domain.Task, error) {
+	return r.Get(ctx, id)
+}
+
+// Update writes the mutated entity back under optimistic concurrency.
+func (r *TaskRepo) Update(ctx context.Context, t *domain.Task) error {
+	res, err := conn(ctx, r.db).ExecContext(ctx, `
+UPDATE tasks SET
+	status = ?, attempt = ?, lease_owner = ?, lease_expires_at = ?,
+	result_artifact_id = ?, metrics = ?, error_code = ?, error_message = ?,
+	started_at = ?, completed_at = ?, version = ?
+WHERE id = ? AND version = ?`,
+		string(t.Status), t.Attempt, nullableString(t.LeaseOwner), encodeTimePtr(t.LeaseExpiresAt),
+		nullableUUID(t.ResultArtifactID), nullableMetrics(t.Metrics),
+		nullableString(t.ErrorCode), nullableString(t.ErrorMessage),
+		encodeTimePtr(t.StartedAt), encodeTimePtr(t.CompletedAt), t.Version,
+		t.ID.String(), t.Version-1)
+	if err != nil {
+		return err
+	}
+	affected, err := res.RowsAffected()
+	if err != nil {
+		return err
+	}
+	if affected == 0 {
+		return domain.ErrLeaseConflict
+	}
+	return nil
+}
+
+func (r *TaskRepo) InsertBatch(ctx context.Context, tasks []*domain.Task) error {
+	for _, t := range tasks {
+		_, err := conn(ctx, r.db).ExecContext(ctx, `
+INSERT INTO tasks (id, job_id, chunk_index, workload, input_uri, input_artifact_id,
+	input_sha256, parameters, status, attempt, max_attempts, created_at, version)
+VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
+			t.ID.String(), t.JobID.String(), t.ChunkIndex, t.Workload,
+			nullIfEmpty(t.InputURI), nullableUUID(t.InputArtifactID),
+			t.InputSHA256, encodeJSON(t.Parameters), string(t.Status),
+			t.Attempt, t.MaxAttempts, encodeTime(t.CreatedAt), t.Version)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (r *TaskRepo) ListCompleted(ctx context.Context, jobID uuid.UUID) ([]*domain.Task, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+taskColumns+" FROM tasks WHERE job_id = ? AND status = ? ORDER BY chunk_index",
+		jobID.String(), string(domain.TaskCompleted))
+	if err != nil {
+		return nil, err
+	}
+	defer func() { _ = rows.Close() }()
+	var tasks []*domain.Task
+	for rows.Next() {
+		task, err := scanTask(rows)
+		if err != nil {
+			return nil, err
+		}
+		tasks = append(tasks, task)
+	}
+	return tasks, rows.Err()
+}
+
+func (r *TaskRepo) CountByStatus(ctx context.Context, jobID uuid.UUID) (map[domain.TaskStatus]int, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT status, count(*) FROM tasks WHERE job_id = ? GROUP BY status", jobID.String())
+	if err != nil {
+		return nil, err
+	}
+	defer func() { _ = rows.Close() }()
+	counts := make(map[domain.TaskStatus]int)
+	for rows.Next() {
+		var (
+			status string
+			n      int
+		)
+		if err := rows.Scan(&status, &n); err != nil {
+			return nil, err
+		}
+		counts[domain.TaskStatus(status)] = n
+	}
+	return counts, rows.Err()
+}
+
+func (r *TaskRepo) CancelByJob(ctx context.Context, jobID uuid.UUID, now time.Time) (int64, error) {
+	res, err := conn(ctx, r.db).ExecContext(ctx, `
+UPDATE tasks SET
+	status = 'cancelled',
+	lease_owner = NULL,
+	lease_expires_at = NULL,
+	error_code = NULL,
+	error_message = NULL,
+	completed_at = ?,
+	version = version + 1
+WHERE job_id = ? AND status IN ('pending','leased','running')`,
+		now.UnixNano(), jobID.String())
+	if err != nil {
+		return 0, err
+	}
+	return res.RowsAffected()
+}
+
+// ExpireLeases applies the lease-expiry rule set-based and returns the
+// distinct jobs whose aggregate status may have changed.
+func (r *TaskRepo) ExpireLeases(ctx context.Context, now time.Time) ([]uuid.UUID, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx, `
+UPDATE tasks SET
+	status = CASE WHEN attempt < max_attempts THEN 'pending' ELSE 'failed' END,
+	lease_owner = NULL,
+	lease_expires_at = NULL,
+	error_code = CASE WHEN attempt >= max_attempts THEN ? ELSE error_code END,
+	error_message = CASE WHEN attempt >= max_attempts THEN ? ELSE error_message END,
+	completed_at = CASE WHEN attempt >= max_attempts THEN ? ELSE completed_at END,
+	version = version + 1
+WHERE status IN ('leased','running') AND lease_expires_at < ?
+RETURNING job_id`,
+		domain.ErrCodeLeaseExpired, "lease expired after the final attempt", now.UnixNano(), now.UnixNano())
+	if err != nil {
+		return nil, err
+	}
+	defer func() { _ = rows.Close() }()
+	var affected []uuid.UUID
+	seen := map[uuid.UUID]bool{}
+	for rows.Next() {
+		var raw string
+		if err := rows.Scan(&raw); err != nil {
+			return nil, err
+		}
+		if id, err := uuid.Parse(raw); err == nil && !seen[id] {
+			seen[id] = true
+			affected = append(affected, id)
+		}
+	}
+	return affected, rows.Err()
+}
+
+// nullableString renders a nilable string, or NULL.
+func nullableString(s *string) any {
+	if s == nil {
+		return nil
+	}
+	return *s
+}
+
+// nullableMetrics stores nil metrics as NULL, else JSON text.
+func nullableMetrics(m map[string]any) any {
+	if m == nil {
+		return nil
+	}
+	return encodeJSON(m)
+}
@@ -0,0 +1,41 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"time"
+
+	"github.com/google/uuid"
+)
+
+// TaskResultRepo records and tallies quorum votes for untrusted task results.
+type TaskResultRepo struct {
+	db *sql.DB
+}
+
+func NewTaskResultRepo(db *sql.DB) *TaskResultRepo {
+	return &TaskResultRepo{db: db}
+}
+
+// RecordVote stores (or replaces) one owner's vote for a task's result.
+func (r *TaskResultRepo) RecordVote(ctx context.Context, taskID, ownerID uuid.UUID, sha256 string, artifactID uuid.UUID) error {
+	_, err := conn(ctx, r.db).ExecContext(ctx, `
+INSERT INTO task_results (task_id, owner_id, result_sha256, result_artifact_id, created_at)
+VALUES (?, ?, ?, ?, ?)
+ON CONFLICT (task_id, owner_id) DO UPDATE SET
+	result_sha256 = excluded.result_sha256,
+	result_artifact_id = excluded.result_artifact_id,
+	created_at = excluded.created_at`,
+		taskID.String(), ownerID.String(), sha256, artifactID.String(), time.Now().UnixNano())
+	return err
+}
+
+// CountAgreeing returns how many distinct owners have voted for the given
+// result hash on this task.
+func (r *TaskResultRepo) CountAgreeing(ctx context.Context, taskID uuid.UUID, sha256 string) (int, error) {
+	var n int
+	err := conn(ctx, r.db).QueryRowContext(ctx,
+		"SELECT count(DISTINCT owner_id) FROM task_results WHERE task_id = ? AND result_sha256 = ?",
+		taskID.String(), sha256).Scan(&n)
+	return n, err
+}
@@ -0,0 +1,78 @@
+package sqlite
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+
+	"github.com/google/uuid"
+
+	"github.com/emil28092005/SciMesh/coordinator/internal/domain"
+)
+
+// UIReadRepo contains bounded, deterministic read queries for the operator UI.
+type UIReadRepo struct{ db *sql.DB }
+
+func NewUIReadRepo(db *sql.DB) *UIReadRepo { return &UIReadRepo{db: db} }
+
+func (r *UIReadRepo) GetJob(ctx context.Context, id uuid.UUID) (*domain.Job, error) {
+	return NewJobRepo(r.db).Get(ctx, id)
+}
+
+func (r *UIReadRepo) ListTasksByJob(ctx context.Context, jobID uuid.UUID) ([]domain.Task, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+taskColumns+" FROM tasks WHERE job_id = ? ORDER BY chunk_index ASC", jobID.String())
+	if err != nil {
+		return nil, fmt.Errorf("list tasks: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	tasks := make([]domain.Task, 0)
+	for rows.Next() {
+		task, err := scanTask(rows)
+		if err != nil {
+			return nil, err
+		}
+		tasks = append(tasks, *task)
+	}
+	return tasks, rows.Err()
+}
+
+func (r *UIReadRepo) ListWorkers(ctx context.Context, limit int) ([]domain.Worker, error) {
+	if limit < 1 || limit > 100 {
+		return nil, domain.ErrInvalidInput
+	}
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+workerColumns+" FROM workers ORDER BY last_heartbeat_at DESC, id DESC LIMIT ?", limit)
+	if err != nil {
+		return nil, fmt.Errorf("list workers: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	workers := make([]domain.Worker, 0)
+	for rows.Next() {
+		worker, err := scanWorker(rows)
+		if err != nil {
+			return nil, err
+		}
+		workers = append(workers, *worker)
+	}
+	return workers, rows.Err()
+}
+
+func (r *UIReadRepo) ListArtifactsByJob(ctx context.Context, jobID uuid.UUID) ([]domain.Artifact, error) {
+	rows, err := conn(ctx, r.db).QueryContext(ctx,
+		"SELECT "+artifactColumns+" FROM artifacts WHERE job_id = ? ORDER BY created_at ASC, id ASC",
+		jobID.String())
+	if err != nil {
+		return nil, fmt.Errorf("list artifacts: %w", err)
+	}
+	defer func() { _ = rows.Close() }()
+	artifacts := make([]domain.Artifact, 0)
+	for rows.Next() {
+		artifact, err := scanArtifact(rows)
+		if err != nil {
+			return nil, err
+		}
+		artifacts = append(artifacts, *artifact)
+	}
+	return artifacts, rows.Err()
+}
--- a/Show More
+++ b/Show More