codebase-indexer

Automatically generate documentation by running an initial codebase scan to create a living docs index. This builds five files covering architecture, implementation, patterns, decisions, and changelog that Claude reads automatically in future sessions instead of rescanning the codebase.

Keep project documentation updated automatically through CLAUDE.md rules installed during the initial scan. These rules trigger documentation refreshes after features or bugfixes, so you never need to manually re-run the indexing skill to maintain your docs index.

Automated codebase indexing works for polyglot monorepos and multiple project types including Node.js, Java, Go, Python, Rust, .NET, and PHP. The scan handles mixed technology stacks and generates a unified documentation index across the entire repository.

Prevent generated documentation files from cluttering git history through automatic .gitignore integration. The skill adds the docs directory to your gitignore file, keeping these session artifacts out of version control while remaining accessible to Claude.

The best way to index an existing codebase is to open your project and say index this codebase. This triggers a comprehensive scan that writes five documentation files under docs and installs CLAUDE.md rules for automatic future updates.

Re-index or refresh project documentation by saying update docs or re-index to Claude. This triggers a new scan to regenerate the architecture, implementation, patterns, decisions, and changelog files under the docs directory.