Indexes and documents codebases for persistent Claude context with minimal token overhead using tiered memory.
Repo Indexer optimizes how Claude understands large codebases by implementing a tiered memory architecture that balances deep technical detail with extreme token efficiency. It systematically analyzes project configurations, entry points, and module structures to generate a lightweight CLAUDE.md file and supplemental documentation in a dedicated .claude/ directory, ensuring Claude retains a high-level overview of the project without exhausting the context window during daily development tasks.
Key Features
013 GitHub stars
02On-demand deep context loading via specialized architecture and convention files
03Automated repository type detection for Monorepos, Libraries, and Microservices
04Tiered memory system (L0-L3) for efficient token budget management
05Strict token validation to keep CLAUDE.md under a 500-token limit
06Systematic 9-point codebase analysis including APIs, data layers, and dependencies
Use Cases
01Bootstrapping persistent project memory for long-term Claude Code sessions
02Documenting architectural patterns and coding conventions for AI consistency
03Onboarding to a new, complex codebase quickly and efficiently
