scheidydude/codeindex

Codeindex is a command-line tool that analyzes how the files in a code repository depend on each other, then uses that dependency map to calculate a "blast radius" score for each file. The blast radius score tells you how many other files would be affected if you changed a particular file. A file that many other parts of the codebase import from will have a high blast score, a file that nothing else depends on will have a low one. This is intended to help developers, and AI coding assistants, understand the risk of making a change before they make it. The tool works with over a dozen programming languages including Python, JavaScript, TypeScript, Go, Ruby, Rust, Java, PHP, and others. It also analyzes Docker service dependencies, GitHub Actions workflow step dependencies, and database schema foreign key relationships. Running the analysis writes a file called codeindex.json into the project, which records every file, what it imports, what imports it, its blast score, and its size in lines of code. A separate command builds a symbol index, which is a map from every function, class, and type name to the exact file and line where it is defined. The README notes that AI coding tools normally find a function by scanning every file in a project, which uses a lot of tokens. With a symbol index, the AI can do a single lookup instead. There is an option to append a compressed version of this symbol map into a CLAUDE.md file so that Claude Code loads it automatically at the start of every session. The tool can also install a git pre-commit hook that checks the blast score of any file you are about to commit and either warns or blocks the commit if the score is above a threshold you set. An optional visualization server renders the dependency graph in a browser as a 2D force layout, 3D network, dependency matrix, or treemap. For AI tool integration specifically, codeindex can run as an MCP server, which is a protocol that lets AI assistants like Claude call it as a tool directly during a conversation. The installation is a single pip command and the tool has no external dependencies beyond the Python standard library.