Codebase Memory

•
49k
•

Indexes codebases into a persistent knowledge graph for efficient, structural exploration and analysis by AI assistants and developers.

Codebase Memory transforms your project's source code into a detailed, persistent knowledge graph, offering a highly efficient alternative to traditional file-by-file exploration. It meticulously parses 35 programming languages to extract functions, classes, modules, call relationships, and cross-service HTTP links using tree-sitter. This structural analysis backend drastically reduces token consumption for AI assistants by providing precise, context-rich results through Cypher-like queries, seamlessly integrating with MCP-compatible clients like Claude Code, Cursor, and Codex CLI, or via its own powerful CLI mode.

主要功能

01Supports 35 programming languages with tree-sitter parsing.
02Generates comprehensive architecture overviews and call graphs.
03Maps Git diffs to affected symbols with risk classification.
04Achieves 99.2% token reduction compared to file-by-file exploration.
05Offers sub-millisecond graph queries and incremental re-indexing.
0683 GitHub stars

使用案例

01Accelerating AI code exploration and understanding for developers.
02Analyzing codebase architecture, dependencies, and hotspots.
03Identifying dead code and assessing the impact of code changes.