files cant't be seen by LLM, but everything else works.

backend/.dockerignore

@@ -0,0 +1,4 @@
+node_modules
+.env
+*.log
+data

backend/Dockerfile

@@ -0,0 +1,20 @@
+FROM node:20-alpine
+
+WORKDIR /app
+
+# Install dependencies
+COPY package*.json ./
+RUN npm ci --only=production
+
+# Copy source
+COPY src/ ./src/
+
+# Create data directories
+RUN mkdir -p /app/data/uploads /app/data/chromadb
+
+EXPOSE 3002
+
+ENV NODE_ENV=production
+ENV PORT=3002
+
+CMD ["node", "src/index.js"]

backend/src/index.js

@@ -6,7 +6,7 @@ const fs = require('fs');
 const path = require('path');
 
 const app = express();
-const PORT = process.env.PORT || 3001;
+const PORT = process.env.PORT || 3002;
 
 // Middleware
 app.use(cors());
@@ -31,6 +31,42 @@ const knowledgeBases = new Map();
 const knowledgeFiles = new Map();
 const uploadedFiles = new Map();
 
+// Token usage tracking
+const tokenUsageStats = new Map(); // conversationId -> aggregated token stats
+
+// In-memory storage for token usage events
+const tokenUsageLog = [];
+
+// Extract token usage from SSE response
+function extractTokenUsage(fullResponse) {
+  try {
+    // Parse the last SSE data block which contains usage info
+    const lines = fullResponse.split('\n');
+    for (let i = lines.length - 1; i >= 0; i--) {
+      const line = lines[i];
+      if (line.startsWith('data: ')) {
+        const data = line.slice(6).trim();
+        if (data === '[DONE]') continue;
+        try {
+          const parsed = JSON.parse(data);
+          if (parsed.usage) {
+            return {
+              promptTokens: parsed.usage.prompt_tokens || 0,
+              completionTokens: parsed.usage.completion_tokens || 0,
+              totalTokens: parsed.usage.total_tokens || 0,
+            };
+          }
+        } catch {
+          // Not valid JSON, skip
+        }
+      }
+    }
+    return null;
+  } catch {
+    return null;
+  }
+}
+
 // Global ChromaDB client (recreated per request when path changes)
 function getChromaClient(persistDir = './data/chromadb') {
   return new ChromaClient({
@@ -78,13 +114,31 @@ app.post('/api/v1/chat/completions', async (req, res) => {
 
     const reader = response.body.getReader();
     const decoder = new TextDecoder();
+    let fullResponse = '';
 
     while (true) {
       const { done, value } = await reader.read();
       if (done) break;
-      res.write(decoder.decode(value, { stream: true }));
+      const chunk = decoder.decode(value, { stream: true });
+      fullResponse += chunk;
+      res.write(chunk);
     }
     res.end();
+
+    // Parse token usage from the final SSE message
+    const tokenUsage = extractTokenUsage(fullResponse);
+    if (tokenUsage) {
+      const usageEntry = {
+        timestamp: Date.now(),
+        conversationId: req.body.conversationId || 'unknown',
+        model,
+        promptTokens: tokenUsage.promptTokens,
+        completionTokens: tokenUsage.completionTokens,
+        totalTokens: tokenUsage.totalTokens,
+      };
+      tokenUsageLog.push(usageEntry);
+      console.log(`Token usage: ${usageEntry.totalTokens} tokens (${usageEntry.promptTokens} prompt, ${usageEntry.completionTokens} completion) for model ${model}`);
+    }
   } catch (error) {
     console.error('Proxy error:', error);
     res.status(500).json({
@@ -427,6 +481,60 @@ app.get('/api/health', (req, res) => {
   res.json({ status: 'ok', timestamp: new Date().toISOString() });
 });
 
+// Token usage tracking endpoints
+app.get('/api/token-usage', (req, res) => {
+  try {
+    const { period = 'all', conversationId } = req.query;
+
+    let filtered = [...tokenUsageLog];
+
+    if (conversationId) {
+      filtered = filtered.filter((e) => e.conversationId === conversationId);
+    }
+
+    if (period === 'today') {
+      const todayStart = new Date();
+      todayStart.setHours(0, 0, 0, 0);
+      filtered = filtered.filter((e) => e.timestamp >= todayStart.getTime());
+    } else if (period === 'week') {
+      const weekStart = new Date();
+      weekStart.setDate(weekStart.getDate() - 7);
+      filtered = filtered.filter((e) => e.timestamp >= weekStart.getTime());
+    }
+
+    const promptTokens = filtered.reduce((sum, e) => sum + e.promptTokens, 0);
+    const completionTokens = filtered.reduce((sum, e) => sum + e.completionTokens, 0);
+    const totalTokens = filtered.reduce((sum, e) => sum + e.totalTokens, 0);
+
+    // Per-model breakdown
+    const modelBreakdown = {};
+    filtered.forEach((entry) => {
+      if (!modelBreakdown[entry.model]) {
+        modelBreakdown[entry.model] = { promptTokens: 0, completionTokens: 0, totalTokens: 0, requestCount: 0 };
+      }
+      modelBreakdown[entry.model].promptTokens += entry.promptTokens;
+      modelBreakdown[entry.model].completionTokens += entry.completionTokens;
+      modelBreakdown[entry.model].totalTokens += entry.totalTokens;
+      modelBreakdown[entry.model].requestCount += 1;
+    });
+
+    res.json({
+      summary: { promptTokens, completionTokens, totalTokens, requestCount: filtered.length },
+      modelBreakdown,
+      history: filtered.slice(-100), // last 100 entries
+    });
+  } catch (error) {
+    console.error('Token usage error:', error);
+    res.status(500).json({ error: error.message });
+  }
+});
+
+app.delete('/api/token-usage', (req, res) => {
+  tokenUsageLog.length = 0;
+  tokenUsageStats.clear();
+  res.json({ success: true });
+});
+
 app.listen(PORT, () => {
   console.log(`AIUI Backend running on port ${PORT}`);
   console.log(`Upload directory: ${uploadDir}`);