API

Use Repokit through direct HTTP requests.

Use the API when you want explicit HTTP + JSON control, straightforward backend integration, and a clean retrieval layer you can embed in tools or workflows.

Quick actions

The API is the most direct way to verify Repokit on your own repository: issue a verification token, send a task, and inspect the ranked file shortlist.

Base URL

https://api.repokit.live

Open

Primary endpoint

https://api.repokit.live/find_relevant_files

Feedback endpoint

https://api.repokit.live/find_relevant_files/feedback

What this page is for

Repokit is repository-aware retrieval. The API gives you direct programmatic access to that retrieval layer once your repository is ready.

The normal beta flow is still repo-first: activate your repository, wait until it reaches ready, then issue a verification token from the control plane.

What you can do today

  • List the repositories enabled for your token.
  • Inspect presets for one repository.
  • Call find_relevant_files for one ready repository.
  • Submit lightweight feedback tied to a request_id.

Fastest verification path

  • Activate a supported repository on submit.repokit.live/control-plane.
  • Wait until the stage bar reaches ready.
  • Issue a verification token from the submission detail.
  • Call find_relevant_files against your repository_id.
  • Compare the ranked shortlist with the files you actually inspect.

Quick request example

find_relevant_filesbash
curl -sS \
  -H "Authorization: Bearer <your_token>" \
  -H "Content-Type: application/json" \
  https://api.repokit.live/find_relevant_files \
  -d '{
    "repository_id": "<repository_id>",
    "query": "fix request validation error in routing",
    "active_files": ["app/routing.py"],
    "recent_files": ["tests/test_routing.py"],
    "failing_tests": ["tests/test_routing.py"],
    "top_k": 5
  }'

What a response gives you

  • A request_id you can keep for feedback or debugging.
  • A ranked files list with scores, confidence, and explanations.
  • Runtime metadata that tells you what served the result.
  • Latency and candidate-universe details for inspection.

When to use the API

  • You want direct HTTP requests rather than a tool protocol.
  • You are integrating retrieval into backend services, editors, or internal tools.
  • You want structured request and response handling with minimal abstraction.