tree-sitting

AST-powered code navigation using tree-sitter. Parses all source files in a codebase into in-memory syntax trees, then provides fast query tools for symbol search, file/directory overview, source retrieval, and reference finding.

Features

• Fast scanning — parses ~250 files in ~700ms, then all queries are sub-millisecond from cache
• Symbol search — find by exact name, substring, or glob pattern across the entire codebase
• Directory and file overview — structural summaries with symbol counts, signatures, and doc comments
• Source retrieval — fetch implementation of any symbol, preferring definitions over declarations
• Reference finding — locate all textual references to a symbol via fast grep against cached source
• 12 bundled grammars — Python, JavaScript, TypeScript, TSX, Go, Rust, Ruby, Java, C, HTML, Markdown, Mojo (extensible: drop more .so files in parsers/)
• Three-tier extraction — custom extractors (richest), community tags.scm queries, and generic heuristic fallback
• Dual deployment — direct Python calls in Claude.ai, or long-lived MCP server in Claude Code

Dependencies

• tree-sitter — bare parser runtime, loads grammars from bundled parsers/*.so
• fastmcp — required only for MCP server mode (Claude Code)

Skill folder: tree-sitting

Release of tree-sitting version 0.7.0

📥 Download & Install

⬇️ Download tree-sitting.zip

To install:

1. Click the download link above (ignore the "Source code" archives below - they're auto-generated by GitHub)
2. Go to Claude.ai Skills Settings
3. Upload the downloaded ZIP file
4. Requires paid Claude Pro or Team account

See official documentation for more details.

Recent Changes

62a1e68 feat(tree-sitting): persistent scan-cache (0.7.0) (#732)
b2fe137 tree-sitting: note tree-sitter v0.26.10 is core-only (no Python binding yet) (#719)
fb6e0fb docs: Update CHANGELOG.md for released skills