Yellhorn MCP

🚀 Yellhorn MCP

A Model Context Protocol (MCP) server that offers Gemini 2.5 Pro and OpenAI capabilities, used to prompt Claude Code with the context of the entire codebase for software development tasks. This mode is ideal for defining work to be performed by code assistants (such as Claude Code or other MCP-compatible coding agents) and ensuring that the results meet the original requirements through comprehensive inspection.

🚀 Quick Start

Yellhorn MCP is a powerful server that integrates multiple model capabilities, enabling you to leverage the context of the entire codebase to perform software development tasks.

✨ Features

• Create Work Plans: Generate detailed implementation plans based on prompts, considering the entire codebase. Publish them as GitHub issues and expose them to coding agents in the form of MCP resources.
• Evaluate Code Diffs: Provide a tool to evaluate whether Git diffs align with the work plans in GitHub issues and create sub - issues in the background.
• Access Resources: Implement the standard MCP resource API to provide access to work plans.
• Development Information: Include installation, configuration, usage instructions, and testing directives.

📦 Installation

# Install production dependencies
pip install .

# Install development dependencies
pip install -e ".[dev]"

📚 Documentation

Configuration

Edit the configuration file:

# MCP configuration example
[mcp]
name = "Yellhorn-MCP"
version = "0.2.1"
description = "Yellhorn Model Context Protocol Server"
author = "your_name_here"
author_email = "your_email_here"
url = "https://github.com/your_account/yellhorn-mcp"

[server]
host = "localhost"
port = 8080
workers = 4

Usage Instructions

Tools

• Create Work Plans

  # Create a new work plan
  mcp create-resource yellhorn-mcp --name "Project Name" --description "Project Description"
  
  # Update an existing work plan
  mcp update-resource yellhorn-mcp 123 --description "Updated Description"
  

• Evaluate Code Diffs

  # Trigger evaluation asynchronously
  mcp judge-resource yellhorn-mcp 123 main HEAD
  
  # View evaluation results
  mcp get-resource-judgment yellhorn-mcp 123
  

Resource Access

Yellhorn MCP implements the standard MCP resource API:

# List all work plans
mcp list-resources yellhorn-mcp

# Get a specific work plan
mcp get-resource yellhorn-mcp 123

Development

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Generate coverage report
pytest --cov=yellhorn_mcp --cov-report term-missing

CI/CD

The project uses GitHub Actions for continuous integration and delivery:

• Testing: Automatically run on pull requests and pushes to the main branch.
  • Use flake8 for code inspection.
  • Use black to check code formatting.
  • Use pytest to execute tests.
• Release: Automatically deploy to PyPI when a commit with a version tag is made.
  • The tag must match the version in pyproject.toml (e.g., v0.2.2).
  • Requires a PyPI API token (PYPI_API_TOKEN) stored in the GitHub repository secrets.

Release a New Version

1. Update the version in pyproject.toml and yellhorn_mcp/init.py.
2. Update CHANGELOG.md and add a change log.
3. Commit the changes: git commit -m "Bump version to X.Y.Z"
4. Tag the commit: git tag vX.Y.Z
5. Push the changes and tags: git push && git push --tags

Reference Documentation

• User Guide: Refer to USAGE.md
• Change Log: View CHANGELOG.md

📄 License

MIT