# jgravelle/jcodemunch-mcp

Cut AI token costs 95%+ on code exploration. The leading MCP server for precise, symbol-level GitHub code retrieval via tree-sitter AST. Works with Claude Code, Cursor & any MCP client. 313B+ tokens saved.

Repository: https://github.com/jgravelle/jcodemunch-mcp
Canonical: https://ross.abutalabs.com/products/jcodemunch-mcp
Homepage: https://jcodemunch.com/
Language: Python
License: NOASSERTION
License Family: other
Topics: claude, claude-code, ai-coding, ast, code-intelligence, context-window, cursor, developer-tools, llm, mcp, mcp-server, model-context-protocol, token-optimization, tree-sitter, cline, codex, copilot, gemini-cli, opencode, windsurf
Last push: 2026-08-26T21:51:20+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 14
- inputs: {"age_days": 206, "days_push": 7, "days_rel": 7, "gap_med": 0, "n_releases_24m": 626}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2624, forks 357 (observed 2026-08-28T04:07:04.931643+00:00)

## What it is
jCodeMunch is an MCP server that indexes codebases with tree-sitter AST parsing and serves symbol-level code retrieval to AI agents, cutting token usage by 86-99% versus reading whole files. It works with Claude Code, Cursor, and any MCP-compatible client, storing structured symbol metadata in a local index for byte-precise fetches.

## Use cases
- reduce AI token costs when exploring large codebases
- let Claude Code or Cursor fetch exact function implementations instead of whole files
- index a repository once and query symbols with byte-level precision
- give AI coding agents scoped context bundles of functions, classes, and imports
- stop burning the LLM context window on irrelevant code lines

## When to choose
- your AI agents spend heavily on reading entire source files
- you use MCP-compatible tools like Claude Code, Cursor, or Windsurf and want precise code retrieval
- you want a local-first code index supporting 70+ languages via tree-sitter

## When to avoid
- you need a fully open-source tool for commercial use without a paid license (dual-use licensing, commercial tiers from $99)
- you only occasionally explore small repos where token savings are negligible
- you require a hosted service rather than a local index

## Facets
- artifact type: service
- maturity: active
- function: mcp, search-engine, parser, developer-tools
- domain: developer-tools, large-language-models, apis
- platform: python, cli, cross-platform, self-hosted
- tags: mcp-server, tree-sitter, code-retrieval, token-optimization, ast-parsing, context-window, code-intelligence, claude-code, cursor, ai-agents

## Member repositories
- jgravelle/jcodemunch-mcp (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:04.931643+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:19:53.117670+00:00, confidence not recorded.
  - readme: https://github.com/jgravelle/jcodemunch-mcp (fetched 2026-08-28T04:07:04.931643+00:00, sha 6c8d55cf7143)
  - homepage: https://jcodemunch.com/ (fetched 2026-08-29T10:02:55.979449+00:00, sha f210f7f27c79)
  - site_page: https://jcodemunch.com/descriptions.php (fetched 2026-08-29T10:02:55.982798+00:00, sha 1154c0b067ea)
- Data as of 2026-08-30T08:39:29.467469+00:00.