Codeix

关于

AI coding agents often struggle with inefficiently finding relevant code, wasting tokens on unstructured searches. Codeix solves this by providing agents with a pre-built, structured map of any codebase, allowing precise queries for symbols, references, and callers. It tackles common issues with existing tools like lack of structure, slow re-parsing, and non-shareable indexes by introducing a git-committable, shareable, and composable `.codeindex` format. This enables fast, structured querying across an entire project and its dependencies, including dedicated prose search for comments and docstrings, all designed to give AI agents exactly what they need in a single, efficient tool call.

主要功能

• 2 GitHub stars
• Generates a git-committable, shareable, and composable `.codeindex` for instant access.
• Includes seven MCP query tools for exploring project structure, searching, and analyzing code relationships.
• Provides structured semantic search for symbols, references, and callers optimized for AI agents.
• Offers fast full-text search across code, comments, docstrings, and string literals.
• Supports a wide range of programming languages and extracts embedded scripts from web formats.

使用案例

• Managing and querying code efficiently across monorepos, sibling projects, and git submodules.
• Enabling developers and CI runners to get instant navigation of code and its dependencies without re-indexing.
• Enhancing AI coding agents by providing structured codebase context, reducing token usage and improving accuracy.