jcodemunch-cli 1.0.0

CLI wrapper for jcodemunch-mcp - Token-efficient code exploration

jCodeMunch CLI

A thin CLI wrapper for jcodemunch-mcp — token-efficient code exploration via tree-sitter AST parsing.

Key feature: Uses uvx to always stay up to date with the latest jcodemunch-mcp. No manual updates needed.

Why This Exists

jcodemunch-mcp is MCP-only (no CLI). This wrapper gives you a CLI that works anywhere:

• Sub-agents via exec
• Scripts
• CI/CD pipelines
• Any AI that can run shell commands

Installation

Option 1: pip (recommended)

pip install jcodemunch-cli

Or with uv:

uv pip install jcodemunch-cli

Option 2: Development Install

git clone https://github.com/your-org/jcodemunch-cli.git
cd jcodemunch-cli
pip install -e .

Option 3: Standalone Script

If you just want to run it directly:

# Requires Python 3.10+ and uv
uv run jcodemunch-cli list-tools

Configuration

Environment Variables

Variable	Required	Description
GITHUB_TOKEN	No	GitHub token for private repos
OPENAI_API_KEY	No	API key for AI summaries
OPENAI_API_BASE	No	API base URL (default: https://api.minimax.chat/v1)
ANTHROPIC_API_KEY	No	Anthropic key for Claude summaries
GOOGLE_API_KEY	No	Google key for Gemini summaries

Set up GitHub Token (optional)

# Using your existing gh CLI
export GITHUB_TOKEN=$(gh auth token)

# Or create a personal access token
# https://github.com/settings/tokens
export GITHUB_TOKEN=ghp_your_token_here

Usage

Index a Repository

# Index a GitHub repository
jcodemunch index-repo owner/repo

# Index a local folder
jcodemunch index-folder /path/to/code

Search for Code

# Search symbols by name
jcodemunch search-symbols --repo owner/repo login

# Get exact symbol code
jcodemunch get-symbol --repo owner/repo "src/main.py::UserService.login#method"

# Get all symbols in a file
jcodemunch file-outline --repo owner/repo src/main.py

# Get repo overview
jcodemunch repo-outline owner/repo

# Full-text search
jcodemunch search-text --repo owner/repo "TODO"

List Indexed Repos

jcodemunch list-repos

For AI Agents

OpenClaw Sub-agents

Add to your skill or agent workflow:

# Agent calls via exec
exec → jcodemunch search-symbols --repo owner/repo functionName

Claude Code (MCP)

Use the native MCP integration instead (recommended for Claude Code):

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

Any Other AI

If your AI can run shell commands, it can use this CLI:

Agent: "Find the login function"
→ exec: jcodemunch search-symbols --repo owner/repo login
→ Returns: [list of matching symbols]
→ exec: jcodemunch get-symbol --repo owner/repo "src/auth.py::login#function"
→ Returns: exact function code

Token Savings

Task	Traditional	jCodeMunch
Find a function	~40,000 tokens	~200 tokens
Understand module API	~15,000 tokens	~800 tokens
Explore repo structure	~200,000 tokens	~2,000 tokens

95%+ token reduction on code exploration tasks.

Requirements

• Python 3.10+
• uv (for automatic jcodemunch-mcp updates)
• GitHub token (optional, for private repos)

How It Works

This package is a thin wrapper that:

1. Uses the mcp Python library to handle the MCP protocol
2. Calls uvx jcodemunch-mcp under the hood
3. Passes through environment variables for API keys

Because it uses uvx, you always get the latest jcodemunch-mcp version — no manual updates required.

License

MIT