mcp-codebase-index
by MikeRecognex·★ 51·Score 47
Zero-dependency MCP server indexing 17 tools for codebase navigation with persistent caching and incremental updates.
Overview
mcp-codebase-index is a structural codebase indexer that parses source files into metadata and exposes 18 query tools via MCP. It supports Python, TypeScript/JS, Go, Rust, and C# files with automatic incremental re-indexing in git repositories. The server uses persistent disk caching to eliminate cold-start penalties and can handle large codebases efficiently (tested on CPython with 1.1M lines).
Try asking AI
After installing, here are 6 things you can ask your AI assistant:
When to choose this
Developers working with medium-to-large Python/TypeScript/Go/Rust/C# codebases who need efficient code navigation and impact analysis without paying token costs for entire file reads.
When NOT to choose this
Projects using languages not supported by this indexer, or when you need write access to codebases (this tool is read-only).
Tools this server exposes
12 tools extracted from the READMEget_project_summaryGet file count, packages, top classes/functions
list_filesList indexed files with optional glob filter
get_structure_summaryGet structure of a file or the whole project
get_functionsList functions with name, lines, params
get_classesList classes with name, lines, methods, bases
get_importsList imports with module, names, line
get_function_sourceGet full source of a function/method
get_class_sourceGet full source of a class
find_symbolFind where a symbol is defined (file, line, type)
get_dependenciesGet what a symbol calls/uses
get_dependentsGet what calls/uses a symbol
get_change_impactGet direct + transitive dependents of a symbol
Comparable tools
Installation
pip install "mcp-codebase-index[mcp]"**Claude Desktop Configuration:** Add to claude_desktop_config.json:
{
"mcpServers": {
"codebase-index": {
"command": "mcp-codebase-index",
"env": {
"PROJECT_ROOT": "/path/to/project"
}
}
}
}FAQ
- How does it handle large codebases efficiently?
- It uses persistent disk caching and incremental updates based on git diff, only re-parsing changed files (~1-2ms). On CPython (1.1M lines), index build takes 56s initially but loads from cache in under 1s on subsequent starts.
- Which languages are supported?
- Python (AST parsing), TypeScript/JS, Go, Rust, C# (regex-based), and Markdown/Text files. Other files only provide line counts.
On Hacker News
Recent discussion from the developer community.
- Story by localforthewin · 2026-02-17
Compare mcp-codebase-index with
Last updated · Auto-generated from public README + GitHub signals.