codeindex
A dependency analyzer that builds a temporal knowledge graph of a codebase and scores each file's blast radius, how many other files would break if it changed, so an AI coding assistant can judge the risk of an edit before making it.
codeindex points at a project in any of several supported languages and builds a persistent, incremental SQLite graph store recording dependencies between files, alongside a JSON dependency index written into the repository for backward compatibility. Blast-radius scores estimate how many files would be affected if a given file changed, including historical as-of queries computed from git history without touching the working tree, and a separate symbol index maps every function, class and type to its exact file and line so an assistant can find a symbol without scanning the whole repository.
A hybrid semantic search fuses natural-language queries with keyword and graph expansion over the indexed symbols. The data can be consumed through the CLI, a Markdown report, an MCP server with ten tools, a pre-commit hook, injection into a project's CLAUDE.md file, or an interactive 2D or 3D visualization with a dependency matrix and treemap. It installs with pip and needs no build step, npm, or required runtime dependency beyond SQLite, which is part of the Python standard library.