adrianczuczka/mason

What it does

Mason is a context engineering MCP (Model Context Protocol) server designed to enhance AI agents' understanding of codebases. It automatically generates a CLAUDE.md file from git history and architectural patterns, and maintains a concept map linking features and flows to their corresponding files. This eliminates the need for agents to repeatedly perform grep and glob operations across the codebase.

How to set up

Mason is deployed as an MCP server compatible with Claude and other AI platforms. Installation involves configuring it as an MCP server in your AI tool's settings, connecting it to your git repository, and allowing it to analyze your codebase structure and commit history to generate the initial context files.

Key features

• Automatic CLAUDE.md generation from git history
• Architectural file sampling and analysis
• Concept mapping of features, flows, and file relationships
• Caching of context snapshots to avoid redundant analysis
• Reduces repetitive file searching for AI agents
• Improved context awareness for multi-query interactions