Mason

MCP server that gives AI assistants a persistent codebase map — feature-to-file lookups, change impact analysis, and git history insights.

Mason creates and maintains a reusable concept map of any git-based codebase, eliminating repeated exploration across sessions. It provides feature-to-file lookup tables stored in a snapshot file, change impact analysis using co-change history and test relationships, and git history aggregation to identify hot files and stale code. Works with any programming language. In a 164-file project, Mason achieved an average 36% token savings while maintaining answer quality.