1
Add it to Claude Code
claude mcp add gemini-mcp -- geminimcpAvailable Tools (4)
Once configured, Gemini MCP gives your AI agent access to:
geminiSend a prompt to Gemini and collect the response.PROMPTcdmodelapproval_modeimage_pathcontextallowed_mcp_serverslist_modelsList available models, approval mode, and bridge status.list_sessionsList active ACP sessions.reset_sessionReset a specific or all sessions.session_idTry It Out
After setup, try these prompts with your AI agent:
→Use Gemini to analyze the current project structure and suggest improvements.
→List all active Gemini ACP sessions and reset the current one.
→Run a task using the gemini tool with the auto_edit approval mode.
→Check which Gemini models are currently available and their bridge status.
Prerequisites & system requirements
- An MCP-compatible client (Claude Code, Cursor, Windsurf, Claude Desktop, or Codex)
- Python 3.8+ with pip installed
Alternative installation methods
Manual (uv)
uv tool install --from git+https://github.com/shenyunhuan/gemini_mcp.git geminimcpKeep this setup from going cold
Save the docs, env vars, and workflow around Gemini MCP in Conare so Claude Code, Codex, and Cursor remember it next time.