Gemini MCP Tool provides an MCP server that enables AI assistants like Claude to interact with the Google Gemini CLI, allowing them to leverage Gemini's large context window for analyzing large files and codebases. The server exposes tools that let Claude ask natural language questions to Gemini and offload token-intensive analysis tasks, solving the problem of token limitations when working with large documents or code repositories. It supports the `@` syntax for directing analysis and can be installed via npm or configured in Claude Desktop for seamless integration.
📚 View Full Documentation - Search me!, Examples, FAQ, Troubleshooting, Best Practices
This is a simple Model Context Protocol (MCP) server that allows AI assistants to interact with the Gemini CLI. It enables the AI to leverage the power of Gemini's massive token window for large analysis, especially with large files and codebases using the @ syntax for direction.
Goal: Use Gemini's powerful analysis capabilities directly in Claude Code to save tokens and analyze large files.
On 2026-06-18, Google retired the Gemini CLI for free, Google AI Pro, and Google AI Ultra users (and individual Gemini Code Assist / GitHub-org users). Its successor is the Antigravity CLI (agy). See Google's announcement.
From 2026-06-18 this tool selects the agy backend automatically. If you are on an affected tier, just install agy:
curl -fsSL https://antigravity.google/cli/install.sh | bash # macOS / Linux
Then run agy once to sign in. Nothing else to change.
Enterprise / Standard-license or paid-API-key user? Your Gemini CLI access is unaffected by the retirement. To keep using it, set one env var on the MCP server:
GEMINI_MCP_BACKEND=gemini
| Variable | Purpose |
|---|---|
GEMINI_MCP_BACKEND | gemini or agy/antigravity. Unset uses the date-aware default (agy from 2026-06-18). |
AGY_CLI_PATH | Full path to the agy binary if it isn't on the server's PATH. |
GEMINI_MCP_TIMEOUT | Overall CLI run timeout in minutes (default 45). |
The agy backend is experimental: print mode is Gemini 3.5 Flash-only, replies come from agy's stdout (transcript recovery only as a fallback), and tool execution isn't sandboxed in -p. The tool emits a notice when a requested model/sandbox can't be honored, and surfaces agy's own errors (quota, auth) verbatim. Full analysis: docs/migration/antigravity-cli.md.
Before using this tool, ensure you have:
claude mcp add gemini-cli -- npx -y gemini-mcp-tool
Use -- y instead of -y:
claude mcp add gemini-cli -- npx -- y gemini-mcp-tool
Type /mcp inside Claude Code to verify the gemini-cli MCP is active.
If you already have it configured in Claude Desktop:
"gemini-cli": {
"command": "npx",
"args": ["-y", "gemini-mcp-tool"]
}
claude mcp add-from-claude-desktop
Register the MCP server with your MCP client:
Add this configuration to your Claude Desktop config file:
{
"mcpServers": {
"gemini-cli": {
"command": "npx",
"args": ["-y", "gemini-mcp-tool"]
}
}
}
If you installed globally, use this configuration instead:
{
"mcpServers": {
"gemini-cli": {
"command": "gemini-mcp"
}
}
}
Configuration File Locations:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.json~/.config/claude/claude_desktop_config.jsonAfter updating the configuration, restart your terminal session.
/gemini-cli and commands will populate in Claude Code's interface.ask gemini to analyze @src/main.js and explain what it doesuse gemini to summarize @. the current directoryanalyze @package.json and tell me about dependenciesask gemini to search for the latest tech newsuse gemini to explain div centeringask gemini about best practices for React development related to @file_im_confused_aboutThe sandbox mode allows you to safely test code changes, run scripts, or execute potentially risky operations in an isolated environment.
use gemini sandbox to create and run a Python script that processes dataask gemini to safely test @script.py and explain what it doesuse gemini sandbox to install numpy and create a data visualizationtest this code safely: Create a script that makes HTTP requests to an APIThese tools are designed to be used by the AI assistant.
ask-gemini: Asks Google Gemini for its perspective. Can be used for general questions or complex analysis of files.
prompt (required): The analysis request. Use the @ syntax to include file or directory references (e.g., @src/main.js explain this code) or ask general questions (e.g., Please use a web search to find the latest news stories).model (optional): The Gemini model to use. Defaults to gemini-2.5-pro.sandbox (optional): Set to true to run in sandbox mode for safe code execution.sandbox-test: Safely executes code or commands in Gemini's sandbox environment. Always runs in sandbox mode.
prompt (required): Code testing request (e.g., Create and run a Python script that... or @script.py Run this safely).model (optional): The Gemini model to use.Ping: A simple test tool that echoes back a message.Help: Shows the Gemini CLI help text.You can use these commands directly in Claude Code's interface (compatibility with other clients has not been tested).
prompt (required): The analysis prompt. Use @ syntax to include files (e.g., /analyze prompt:@src/ summarize this directory) or ask general questions (e.g., /analyze prompt:Please use a web search to find the latest news stories).prompt (required): Code testing request (e.g., /sandbox prompt:Create and run a Python script that processes CSV data or /sandbox prompt:@script.py Test this script safely).message (optional): A message to echo back.Contributions are welcome! Please see our Contributing Guidelines for details on how to submit pull requests, report issues, and contribute to the project.
This project is licensed under the MIT License. See the LICENSE file for details.
Disclaimer: This is an unofficial, third-party tool and is not affiliated with, endorsed, or sponsored by Google.
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent