Jcodemunch Mcp

Name: Jcodemunch Mcp
Author: jgravelle

@jgravelle

About 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.

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "jcodemunch": {
      "command": "uvx",
      "args": [
        "jcodemunch-mcp"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is Jcodemunch Mcp?

Jcodemunch Mcp is a token-efficient MCP server for precise GitHub source code retrieval via tree-sitter AST parsing. It cuts AI token costs by 95%+ by indexing codebases once and retrieving only exact symbols (functions, classes, methods) instead of entire files. It is designed for developers using AI agents like Claude Code, Cursor, VS Code, and any MCP-compatible client.

How to use Jcodemunch Mcp?

Install via pip install jcodemunch-mcp or uvx jcodemunch-mcp. One-click install buttons are available for VS Code, VS Code Insiders, and Cursor. A quickstart guide is at QUICKSTART.md, and full documentation covers tool reference, configuration, and agent hooks.

Key features of Jcodemunch Mcp

Precise symbol retrieval (functions, classes, methods, constants)
95%+ token reduction in code-reading workflows
tree-sitter based AST indexing for polyglot parsing
Compact wire format (MUNCH) for additional token savings
BM25 search, fuzzy matching, semantic/hybrid search (opt-in)
Multi-repo integration, dependency indexing, dead code detection

Use cases of Jcodemunch Mcp

Finding a function implementation without scanning entire files
Understanding module structure via outlines and targeted bundles
Exploring repo structure with query-driven context retrieval
Reducing AI token spend in retrieval-heavy workflows
Analyzing code health with observatory tools (e.g., six-axis health snapshots)

FAQ from Jcodemunch Mcp

What does Jcodemunch Mcp do differently than reading entire files?

It indexes the codebase using tree-sitter and retrieves only the exact code needed (symbols, outlines, context bundles) with byte-level precision, cutting token usage by 95%+.

What are the runtime requirements?

Jcodemunch Mcp requires Python and can be installed via pip or uvx. It runs locally and works with any MCP-compatible client (Claude Code, Cursor, VS Code, etc.).

Where does the data live?

Indexes are stored locally on the user's machine. The server is local-first; no external data transfer is required.

What transports and authentication does it use?

It uses the standard MCP protocol (stdio transport by default). No authentication mechanism is specified in the README.

Is Jcodemunch Mcp free for commercial use?

No, it is free for non-commercial use only. Commercial use requires a paid license: Builder ($79/1 dev), Studio ($349/up to 5 devs), or Platform ($1,999/org-wide).