feat: Add Model Context Protocol (MCP) integration

.env.mcp

@@ -0,0 +1,10 @@
+# MCP Server Configuration
+MCP_PORT=8811
+PROMETHEUS_URL=http://prometheus:9090
+GRAFANA_URL=http://grafana:3001
+MODEL_RUNNER_URL=http://model-runner:12434
+JAEGER_URL=http://jaeger:16686
+
+# Docker Network
+COMPOSE_PROJECT_NAME=genai-mcp
+DOCKER_BUILDKIT=1

README.md

@@ -271,3 +271,30 @@ Contributions are welcome! Please feel free to submit a Pull Request.
 3. Commit your changes (`git commit -m 'Add some amazing feature'`)
 4. Push to the branch (`git push origin feature/amazing-feature`)
 5. Open a Pull Request
+
+## 🤖 MCP Integration
+
+This project now includes Model Context Protocol (MCP) integration, enabling AI-powered monitoring through Claude Desktop.
+
+### Features
+- Real-time performance analysis through Claude
+- AI-powered health recommendations
+- Custom Prometheus query execution
+- Intelligent resource optimization suggestions
+
+### Quick Setup
+```bash
+# Build and start MCP integration
+./start-mcp-stack.sh
+
+# Configure Claude Desktop with the provided config
+# Start asking Claude about your model performance!
+```
+
+For detailed MCP documentation, see [docs/mcp/README.md](docs/mcp/README.md).
+
+### Example Claude Interactions
+- "How is my model performing right now?"
+- "Analyze GPU utilization and suggest optimizations"
+- "Show me error patterns from the last hour"
+- "What's the 95th percentile latency trend?"

docker-compose.mcp-override.yml

@@ -0,0 +1,39 @@
+version: '3.8'
+
+services:
+  mcp-metrics-server:
+    build:
+      context: ./mcp-metrics-server
+      dockerfile: Dockerfile
+    container_name: genai-mcp-metrics
+    environment:
+      - PROMETHEUS_URL=http://prometheus:9090
+      - GRAFANA_URL=http://grafana:3001
+      - MODEL_RUNNER_URL=http://model-runner:12434
+      - JAEGER_URL=http://jaeger:16686
+      - MCP_PORT=8811
+    ports:
+      - "8811:8811"
+    depends_on:
+      - prometheus
+      - grafana
+    networks:
+      - default
+    restart: unless-stopped
+    healthcheck:
+      test: ["CMD-SHELL", "nc -z localhost 8811 || exit 1"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+
+  # Optional: Add a dedicated stdio bridge if needed
+  mcp-bridge:
+    build:
+      context: ./mcp-stdio-bridge
+      dockerfile: Dockerfile
+    environment:
+      - TARGET_HOST=mcp-metrics-server
+      - TARGET_PORT=8811
+    depends_on:
+      - mcp-metrics-server
+    restart: unless-stopped

docs/mcp/README.md

@@ -0,0 +1,77 @@
+# GenAI Model Runner MCP Integration
+
+## Overview
+
+This directory contains the Model Context Protocol (MCP) integration for the GenAI Model Runner metrics project. MCP enables AI-powered monitoring and analysis through Claude Desktop.
+
+## Features
+
+- **Real-time Performance Metrics**: Monitor model latency, throughput, and resource usage
+- **AI-Powered Health Analysis**: Get intelligent insights about model health
+- **Custom Prometheus Queries**: Execute and analyze PromQL queries through Claude
+- **Automated Recommendations**: Receive optimization suggestions for performance and resources
+
+## Quick Start
+
+1. **Build the MCP server**:
+   ```bash
+   cd mcp-metrics-server
+   npm install
+   npm run build
+   ```
+
+2. **Start the integrated stack**:
+   ```bash
+   ./start-mcp-stack.sh
+   ```
+
+3. **Configure Claude Desktop**:
+   Add to your Claude Desktop config:
+   ```json
+   {
+     "mcpServers": {
+       "genai-metrics": {
+         "command": "docker",
+         "args": [
+           "run", "-i", "--rm", "--network=host",
+           "alpine/socat", "STDIO", "TCP:localhost:8811"
+         ]
+       }
+     }
+   }
+   ```
+
+4. **Test the connection**:
+   ```bash
+   ./test-mcp-server.sh
+   ```
+
+## Available MCP Tools
+
+### get_model_performance
+Monitor real-time model performance metrics.
+
+**Example**: "Show me the current model performance for the last 15 minutes"
+
+### analyze_model_health
+Get AI-powered health analysis with recommendations.
+
+**Example**: "Analyze my model health and suggest optimizations"
+
+### get_prometheus_query
+Execute custom Prometheus queries.
+
+**Example**: "Run this query: rate(http_requests_total[5m]) and explain the results"
+
+## Architecture
+
+```
+Claude Desktop ←→ Socat Bridge ←→ MCP Server ←→ Prometheus/Grafana
+```
+
+## Support
+
+For issues or questions about MCP integration:
+1. Check the troubleshooting section in the main README
+2. Review Docker logs: `docker logs genai-mcp-metrics`
+3. Test connectivity: `./test-mcp-server.sh`

grafana/dashboards/mcp-metrics-dashboard.json

@@ -0,0 +1,36 @@
+{
+  "dashboard": {
+    "title": "GenAI MCP Metrics Dashboard",
+    "tags": ["mcp", "genai", "metrics"],
+    "timezone": "browser",
+    "panels": [
+      {
+        "title": "Model Performance",
+        "type": "stat",
+        "targets": [
+          {
+            "expr": "histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m]))",
+            "legendFormat": "95th percentile latency"
+          }
+        ],
+        "gridPos": {"h": 8, "w": 12, "x": 0, "y": 0}
+      },
+      {
+        "title": "Request Rate",
+        "type": "graph",
+        "targets": [
+          {
+            "expr": "rate(model_requests_total[5m])",
+            "legendFormat": "Requests/sec"
+          }
+        ],
+        "gridPos": {"h": 8, "w": 12, "x": 12, "y": 0}
+      }
+    ],
+    "time": {
+      "from": "now-1h",
+      "to": "now"
+    },
+    "refresh": "5s"
+  }
+}

mcp-metrics-server/Dockerfile

@@ -0,0 +1,26 @@
+FROM node:18-alpine
+
+WORKDIR /app
+
+# Install dependencies
+COPY package*.json ./
+RUN npm install
+
+# Copy source code
+COPY src/ ./src/
+COPY tsconfig.json ./
+
+# Build TypeScript
+RUN npm run build
+
+# Install socat for TCP server functionality
+RUN apk add --no-cache socat netcat-openbsd
+
+# Create startup script
+RUN echo '#!/bin/sh' > /app/start.sh && \
+    echo 'node dist/index.js | socat STDIO TCP-LISTEN:$MCP_PORT,reuseaddr,fork' >> /app/start.sh && \
+    chmod +x /app/start.sh
+
+EXPOSE 8811
+
+CMD ["/app/start.sh"]

mcp-metrics-server/package.json

@@ -0,0 +1,21 @@
+{
+  "name": "genai-mcp-metrics-server",
+  "version": "1.0.0",
+  "description": "MCP server for GenAI model runner metrics",
+  "main": "dist/index.js",
+  "type": "module",
+  "scripts": {
+    "build": "tsc",
+    "start": "node dist/index.js",
+    "dev": "tsx src/index.ts"
+  },
+  "dependencies": {
+    "@modelcontextprotocol/sdk": "^0.5.0",
+    "axios": "^1.6.0"
+  },
+  "devDependencies": {
+    "@types/node": "^20.0.0",
+    "typescript": "^5.0.0",
+    "tsx": "^4.0.0"
+  }
+}