Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
76 lines (54 loc) · 2.78 KB

CONTRIBUTING.md

File metadata and controls

76 lines (54 loc) · 2.78 KB

Contributing to claude-code-delegate-local

Thanks for considering a contribution! This project is small and pragmatic — issues, PRs, and discussions are all welcome.

Quick start for developers

git clone https://github.com/fegone/claude-code-delegate-local.git
cd claude-code-delegate-local
uv sync

Run a smoke test against a local backend:

DELEGATE_LOCAL_KEY=your-key uv run python server.py
# starts MCP on stdio — Ctrl+C to stop

Test the tools manually from a Python REPL:

import asyncio, server
server.LITELLM_URL = "http://localhost:4000/v1/messages"
server.LITELLM_KEY = "your-key"
print(asyncio.run(server.local_backend_status()))
print(asyncio.run(server.list_local_agents()))

Submitting changes

  1. Open an issue first if you're proposing a significant change. Avoids wasted work.
  2. Fork the repo, create a feature branch (feat/your-feature or fix/your-bug).
  3. Keep PRs focused — one concern per PR.
  4. Add a CHANGELOG entry under [Unreleased] if your change is user-visible.
  5. If you add a new model prefix to the dual-format routing, document it in docs/CONFIGURATION.md.

Code style

  • Python 3.11+ (matches the .python-version in the repo).
  • Type hints on public functions.
  • httpx for HTTP (already a dep).
  • Keep dependencies minimal — the whole point is a small footprint MCP.
  • Match the existing module layout: helpers grouped by section with comment dividers (# ────...).

Testing changes

No formal test suite yet (PRs welcome to add one). For now, the minimum bar before submitting is:

  1. local_backend_status() returns successfully against your backend.
  2. list_local_agents() returns at least one agent.
  3. delegate_to_local_agent with a tiny task (e.g., "write 'hello world' to /tmp/hi.txt") completes in 1-2 turns without errors.
  4. If you touched the OpenAI-format codepath, repeat #3 with a deepseek-* or openai-* model.

Reporting bugs

Open an issue with:

  • Backend type (LiteLLM proxy / llama.cpp direct / Ollama / DeepSeek direct / etc.)
  • Model name as configured
  • The exact tool call (with secrets redacted) and the full response JSON
  • Server logs if available (uv run python server.py and copy stderr)

For HTTP 4xx errors from the backend, the response body is usually the most informative part.

Things we'd love help with

  • A small pytest-based test suite (mocked backend, no real network calls).
  • Sample agent definitions in examples/agents/ to make first-run easier.
  • A walkthrough video or GIF for the README.
  • Translations of the README to other languages (currently EN/ES).
  • A Docker compose file that bundles LiteLLM + sample local model + this MCP for one-command setup.

License

By contributing you agree your contributions are licensed under the project's MIT license.