Code Context

Compresses codebases into LLM-ready context using AST analysis with surgical read/write tools for TypeScript, Python, PHP, and Dart.

This MCP server extracts structural information from codebases at the symbol level—functions, classes, and methods—rather than sending raw files, achieving 50–80% token reduction. It provides eleven tools split between read operations (semantic repo maps, surgical file reading, impact analysis, line-range reading, pattern search, rollback) and write operations (symbol-level insertion, renaming, removal, and surgical replacement). Uses SQLite caching and supports TypeScript, JavaScript, Python, PHP, and Dart.