gemini-code-assist

Gemini Code Assist MCP integrates the Google Gemini CLI with Claude Code to provide AI-powered development assistance directly within Claude Code. It offers a suite of tools for code analysis, security reviews, feature planning, bug analysis, debugging support, and code understanding. After starting the MCP server, you can invoke tools like gemini_review_code, gemini_analyze_security, gemini_proofread_feature_plan, gemini_suggest_implementation, gemini_analyze_bug, gemini_debug_assistance, gemini_explain_code, and gemini_generate_tests to get structured feedback, recommendations, and explanations tailored to your Python projects. The server relies on your existing Google Cloud authentication (no API key required) and works with both Claude Code and Claude Desktop through the MCP interface.

Prerequisites:

• Python 3.11+ installed on your environment
• UV package manager installed (uv)
• Gemini CLI installed and configured
• Google Cloud authentication configured (gcloud auth login)

Installation steps:

# 1) Clone the MCP server repository
git clone https://github.com/VinnyVanGogh/gemini-code-assist-mcp.git
cd gemini-code-assist-mcp

# 2) Install dependencies (if needed)
# This project uses Python; ensure dependencies are installed via pip if a requirements file exists
# Example: python -m pip install -r requirements.txt

Configure and run the MCP server locally:

# 3) Start the MCP server using UV
uv sync
uv run python src/main.py

Claude Code setup (recommended):

# 4) Install in Claude Code (local MCP server)
uv sync
uv run mcp install src/main.py

# 5) Or add a custom named server via CLI configuration
claude mcp add '{
  "name": "Gemini Code Assist",
  "command": "uv",
  "args": ["run", "python", "src/main.py"],
  "cwd": "/path/to/gemini-code-assist-mcp",
  "env": {
    "GEMINI_DEBUG": "false"
  }
}'

Validation steps:

# 6) Test Gemini CLI access
gemini --help

# 7) Test the MCP server
uv run python src/main.py

Tips and common issues:

• Ensure Gemini CLI is authenticated and available in your PATH; the MCP server relies on Google Cloud authentication, not an API key.
• If you encounter environment or path issues in Claude Code, specify the working directory (cwd) in your MCP configuration.
• Set GEMINI_DEBUG to true during troubleshooting to get verbose logs.
• If the server doesn’t appear in Claude Code, restart Claude Code/Desktop and re-run the MCP sync.
• For remote deployments, you can expose your MCP server URL and connect Claude Code to it using standard mcpServers configuration.