MCP (Model Context Protocol)

The server exposes a Model Context Protocol endpoint for AI assistants and LLM integrations.

POST /mcp

The MCP endpoint is stateless — each request is independent and authenticated via the Authorization header (Bearer token or Basic Auth). There is no server-side session tracking, so clients survive server restarts seamlessly as long as they hold a valid token.

Inspect and search pipeline runs programmatically using MCP tools:

• get_run — fetch run status and metadata
• list_run_tasks — list tasks in a run with outputs
• get_run_task — fetch a single task in a run with full payload/output
• search_tasks — full-text search task outputs
• search_pipelines — search stored pipelines by name/content

Authentication

When the server has authentication enabled, MCP requests require an Authorization header — either Bearer <token> (OAuth) or Basic <base64> (Basic Auth). See Authentication.

Client Setup

See MCP for VS Code extension setup and detailed usage.

Example (Direct HTTP)

curl -X POST http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <token>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "get_run",
      "arguments": { "run_id": "run-123" }
    }
  }'

See MCP for full tool reference and client implementations.

Example (Single Task Full Payload)

curl -X POST http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <token>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "get_run_task",
      "arguments": {
        "run_id": "run-123",
        "path": "jobs/review-pr/1/agent/code-quality-reviewer"
      }
    }
  }'