LGTMCP Server

LGTMCP

A Model Context Protocol (MCP) server that provides AI-powered code review using Google Gemini 3.1 Pro. LGTMCP reviews your code changes and either commits them automatically (if approved) or provides detailed feedback for improvements.

In my usage, the median review takes 1.9 minutes and costs $0.20, with an acceptance rate around 45%. You should decide whether that is slow and expensive or fast and cheap.

Features

• AI Code Review: Leverages Google Gemini 3.1 Pro for intelligent code analysis
• Automatic Commit: Commits changes when code passes review (optional)
• Security Scanning: Built-in secret detection using Gitleaks
• Gitignore Protection: Prevents access to gitignored files during review
• MCP Integration: Works seamlessly with Claude Desktop and other MCP clients
• Review-Only Mode: Option to get feedback without automatic commits

Installation

Build from source

git clone https://msrl.dev/lgtmcp.git
cd lgtmcp
make build

Install to ~/bin

make install

This installs the binary to ~/bin by default. You can customize the installation directory:

make install INSTALL_PATH=/usr/local/bin

Note: Ensure ~/bin is in your shell's PATH. Add this to your shell configuration file if needed:

# For bash/zsh
export PATH="$HOME/bin:$PATH"

Configuration

1. Get a Google API key from Google AI Studio.

2. Create configuration directory:

   mkdir -p ~/.config/lgtmcp
   

3. Create configuration file from example:

   cp config.example.yaml ~/.config/lgtmcp/config.yaml
   

4. Edit the configuration file with your settings:

   google:
     api_key: "your-gemini-api-key-here"
   gemini:
     model: "gemini-3.1-pro-preview"
     fallback_model: "gemini-2.5-pro" # Default; set to "none" to disable
   logging:
     level: "info"
   

The fallback_model is use when we run into quota exhaustion on the primary model. While Gemini 3.1 Pro is in preview, it has very low daily rate limits.

Claude Code configuration

1. Set up configuration file as described above

2. Configure LGTMCP with Claude Code:

claude mcp add lgtmcp -- lgtmcp

Usage

Basic Usage

The MCP server exposes two tools:

`review_only`

Reviews code changes and returns feedback without committing.

Parameters:

• directory: Path to the git repository

`review_and_commit`

Reviews code changes and commits if approved. This is a separate tool so that you can set tool permissions on it differently from review.

Parameters:

• directory: Path to the git repository
• commit_message: Message for the commit if approved

Example Workflows

Review only (no commit):

review_only("/path/to/repo")

Review and commit if approved:

review_and_commit("/path/to/repo", "Add new feature")

What Happens

1. Security check: Scans files for secrets using Gitleaks
2. Diff generation: Creates diff of all staged and unstaged changes
3. AI review: Sends diff to Gemini 3.1 Pro for analysis
   • Gemini can request file contents for context
   • Gitignored files are automatically blocked from access
4. Decision:
   • If approved (LGTM): Returns approval message (review_only) or commits changes (review_and_commit)
   • If not approved: Returns detailed feedback

Configuration

All configuration is managed through the YAML configuration file located at:

• $XDG_CONFIG_HOME/lgtmcp/config.yaml (if XDG_CONFIG_HOME is set)
• ~/.config/lgtmcp/config.yaml (default)

See config.example.yaml for all available configuration options.

Logging

LGTMCP logs are written to platform-specific default locations:

• macOS: ~/Library/Logs/lgtmcp/lgtmcp.log
• Linux: ~/.local/share/lgtmcp/logs/lgtmcp.log (or $XDG_DATA_HOME/lgtmcp/logs/lgtmcp.log)
• Windows: %LOCALAPPDATA%\lgtmcp\logs\lgtmcp.log

You can configure logging in your config.yaml:

logging:
  output: "directory" # Options: none, stdout, stderr, directory, mcp
  level: "info" # Options: debug, info, warn, error
  # directory: "/custom/log/path"  # Optional custom directory

To view logs on macOS:

# View the log file
tail -f ~/Library/Logs/lgtmcp/lgtmcp.log

# Or open in Console.app
open ~/Library/Logs/lgtmcp/lgtmcp.log

Development