svg153/github-stars-contrib-mcp-server

GitHub Stars Contributions MCP Server

Model Context Protocol (MCP) server for the GitHub Stars program: https://stars.github.com/.

It provides a comprehensive API to manage contributions, links, and profile data using your personal API token via the Stars GraphQL API.

• Transport: FastMCP (SSE streamable-http)
• Features: Full CRUD operations for contributions and links, profile management

Configuration

Environment variables:

• STARS_API_TOKEN: Token shown in GitHub Stars profile (Contributions → Token)
• LOG_LEVEL: DEBUG|INFO|WARNING|ERROR|CRITICAL (default: INFO)
• MCP_HOST: Bind host for SSE transport (default: 127.0.0.1)
• MCP_PORT: Bind port (default: 8766)
• MCP_PATH: Path (default: /mcp)
• MCP_TRANSPORT: Transport selection ("stdio" | "http" | "streamable-http" | "sse"). Default: stdio.
• VALIDATE_URLS: Enable lightweight URL HEAD validation before mutations (default: false)
• URL_VALIDATION_TIMEOUT_S: Timeout for URL validation requests (default: 3)
• URL_VALIDATION_TTL_S: TTL cache in seconds for URL validation results (default: 3600)

Tools

The server provides the following MCP tools:

Mutations (write operations):

• create_contribution: Create a single contribution
• create_contributions: Create one or more contributions in a single mutation
• update_contributions: Update one or more contributions
• delete_contributions: Delete one or more contributions
• create_link: Create a link in GitHub Stars profile
• update_link: Update a link in GitHub Stars profile
• delete_link: Delete a link from GitHub Stars profile
• update_profile: Update the user's profile information

Queries (read operations):

• get_user: Get the logged user (minimal user info without nominee data)
• get_stars: Get the public profile stars/contributions for a GitHub user
• search_contributions: Filter contributions from a user's public Stars profile
• get_contributions_stats: Get contribution statistics and analytics (groupable by type, month, year)
• compare_contributions: Compare contributions between two users
• export_contributions: Export contributions in JSON, CSV, or Markdown format

All tools return a standardized response format: { "success": boolean, "data": object | null, "error": string | null }

Quick start

1. Create a virtualenv and install

   python3 -m venv .venv
   source .venv/bin/activate
   python -m pip install -e .[dev]
   

2. Configure environment (from template)

   cp env.template .env
   # edit .env and set STARS_API_TOKEN
   set -a; source .env; set +a
   

   Alternatively, export variables directly:

   export STARS_API_TOKEN="<your_token>"
   export LOG_LEVEL=INFO
   

3. Run the MCP server (Streamable HTTP)

   python -m github_stars_contrib_mcp.server
   # Binds to MCP_HOST (default 127.0.0.1) and MCP_PORT (default 8766), path /mcp
   

Local usage without an MCP client

You can test the tools locally via a tiny demo script. First, make sure a Stars API token is available (export it or put it in a .env file at the project root as STARS_API_TOKEN=...).

export STARS_API_TOKEN=stars_...your_token...

Then run:

python scripts/demo.py get-user-data
python scripts/demo.py create-contributions --data '[{"title": "Test Contribution", "url": "https://example.com", "type": "BLOGPOST", "date": "2023-01-01T00:00:00Z"}]'

Shortcut with Makefile (loads .env, activates venv if present):

make run-env

Tool contracts are defined in src/github_stars_contrib_mcp/tools/ (source of truth).

Enums and aliases

For link platforms, see PlatformType in src/github_stars_contrib_mcp/models.py. The tools also accept a couple of legacy aliases which are normalized internally by normalize_platform in src/github_stars_contrib_mcp/utils/normalization.py (e.g., GITHUB → README, WEBSITE → OTHER). When an alias is used, a warning is logged to help migrate callers.

Testing

Unit tests:

. .venv/bin/activate
pytest -q

All tests, with optional live mutation e2e:

# Create `.env.local` with at least STARS_API_TOKEN; set STARS_E2E_MUTATE=1 to enable mutation tests
make test-all

Notes:

• If your token doesn’t expose nominee data, read/profile flows may be skipped by tests (that’s expected).
• Ensure LOG_LEVEL is one of: DEBUG, INFO, WARNING, ERROR, CRITICAL.

Features and Roadmap

This project includes comprehensive improvements planned across 10 key areas:

• Security and Authentication
• Error Handling System
• Data Validation
• Performance Optimizations
• Complete Documentation
• CI/CD Automation
• Monitoring and Observability
• Advanced Testing
• Advanced Features
• Advanced Architecture

See docs/feats/README.md for detailed implementation plans and phased rollout strategy.