chat: mission mode (/mission, --mission) — autonomous goal runs in background with retries, corrections, state-aware retry hints; input() moved to thread so missions run while the REPL waits; configurable nudge_limit

testbed/llm_agent.py

@@ -363,6 +363,7 @@ async def run_llm_agent_loop(
     *,
     log: LogFn,
     recorder: Any | None = None,
+    nudge_limit: int = 1,
 ) -> dict[str, Any]:
     """Drive the controller until the mission is done or steps run out."""
     summary: dict[str, Any] = {
@@ -450,11 +451,13 @@ async def run_llm_agent_loop(
         if not turn.calls:
             # Nudge a model that drifted into plain text back to acting.
             silent = (not turn.text) or len(turn.text) < 8
-            already_nudged = (
-                bool(controller.messages)
-                and controller.messages[-1].get("content") == NUDGE
-            )
-            if silent or already_nudged:
+            consecutive_nudges = 0
+            for msg in reversed(controller.messages):
+                if msg.get("content") == NUDGE:
+                    consecutive_nudges += 1
+                else:
+                    break
+            if silent or consecutive_nudges >= nudge_limit:
                 summary["result"] = "model produced no tool call"
                 return summary
             controller.messages.append({"role": "user", "content": NUDGE})