Index Java, Go, Python, and Rust codebases into graph.json and MAP.md.
npx skills add https://github.com/kenlck/skills --skill codemap-kenlck
Please help me install this Agent Skill. Skill: codemap Source: https://github.com/kenlck/skills/tree/main/ken-swe/skills/codemap Command: npx skills add https://github.com/kenlck/skills --skill codemap-kenlck
Build and maintain a persistent structural index of a multi-language codebase under .codemap/ (graph.json + MAP.md) so teams can explore symbols, imports, and relationships without grep-and-read loops. It supports Java, Go, Python, and Rust, and is designed to be used after large refactors or at the start of exploration in unfamiliar repos; for TypeScript/JavaScript, the sibling ts-codemap offers higher fidelity.
Run codemap build on a repository to generate the graph and MAP.md under .codemap.
High-intent search queries and answers about installing and using this skill.
You can index a polyglot codebase by running codemap build, which uses tree-sitter to parse Java, Go, Python, and Rust source files. It writes a persistent graph.json and MAP.md under .codemap/ to enable fast exploration.
Yes, codemap uses tree-sitter to map codebase structure after major refactors. It surfaces module boundaries, symbol exports, and import/calls edges to guide exploration and planning in Java, Go, Python, and Rust monorepos.
Codemap does not directly support TypeScript and JavaScript codebase navigation. For TypeScript/JavaScript repositories, the sibling ts-codemap skill offers higher fidelity structural indexing and exploration.
Incremental codebase indexing uses per-file SHA256 caching to handle updates efficiently. This allows codemap to perform fast rebuilds by only re-indexing files that have changed, rather than walking the entire repository every time.
The best way to explore unfamiliar code without grep-and-read loops is to generate a persistent structural index. Codemap writes a graph map to .codemap/, allowing you to query symbols, imports, and relationships directly.
You should use tree-sitter for codebase mapping because it parses source code structurally rather than lexically. This extracts accurate symbols, imports, and calls edges into a persistent graph, eliminating the manual grep-and-read exploration loop.
plurigridParse source code into ASTs and extract symbols across multiple languages.
tedtv1007-ctrlQuery a JSON knowledge graph for codebase architecture and dependency mapping.
kenlckBuild a persistent graph of TypeScript/JavaScript files, symbols, and edges.
enigmaicon-engDiscovers and navigates code structure using AST-based search and targeted symbol expansion.
