Control Plane

Repository automation now has a small hosted runtime surface.

RepoDocs AI can now expose validation, export, agent, analytics, and graph jobs through a small Node control plane with HTTP endpoints, Redis-backed queue state, per-user keys, and a lightweight admin dashboard.

Endpoints

Runnable over HTTP

  • GET /health
  • GET /auth/status
  • GET /jobs
  • GET /runs
  • GET /artifacts
  • GET /users
  • POST /jobs/validate
  • POST /jobs/export
  • POST /jobs/automation
  • POST /users and /users/:id/keys

Hosted Scope

Durable and user-managed

The control plane is meant to bridge repo-native automation and a hosted deployment model without introducing a full platform stack. It now protects job and artifact endpoints with per-user API keys, persists queue state and user metadata in Redis for multi-worker execution, exposes a small browser admin surface for user and key management, and can be deployed with the included Render, Fly.io, or Azure Container Apps manifests.

Runtime

Start it locally or run it in a container.

Run npm run control-plane:start with REDIS_URL and REPODOCS_CONTROL_PLANE_BOOTSTRAP_KEY set, or build the container image with npm run docker:control-plane:build and deploy it with the included Render, Fly.io, or Azure Container Apps manifests.