Gemini Mcp Bridge
Give Claude Code or Cursor a second Gemini-powered brain for queries, codebase search, reviews, structured JSON, and diff checks without leaving the agent.
Overview
gemini-mcp-bridge is an MCP server for the Build phase that exposes Gemini CLI capabilities—query, review, search, structured output, and diff assessment—to your coding agent over stdio.
What is this MCP server?
- Wraps Gemini CLI via stdio MCP with configurable GEMINI_CLI_PATH and default/fallback models
- Supports query, review, search, structured output, and diff assessment flows
- Limits load with GEMINI_MAX_CONCURRENT (default 3) and GEMINI_MIN_INVOCATION_GAP_MS (default 5000)
- Falls back to GEMINI_FALLBACK_MODEL on quota or availability errors
- npm package gemini-mcp-bridge v0.6.0
- Server version 0.6.0; default GEMINI_MAX_CONCURRENT 3; default GEMINI_MIN_INVOCATION_GAP_MS 5000
Community signal: 6 GitHub stars.
What problem does it solve?
You want Gemini’s reasoning and search in the same agent workflow, but juggling a second terminal and manual paste breaks flow and wastes context.
Who is it for?
Indie devs who already use Gemini CLI and want a governed subprocess bridge inside Claude Code or Cursor.
Skip if: Builders who cannot install or authorize Gemini CLI locally, or who only need a hosted chat UI with no MCP host.
What do I get? / Deliverables
After you register the server, your agent can call Gemini-backed tools with shared rate limits and model fallback while you stay in one repo session.
- Callable MCP tools that run Gemini CLI for query, review, search, structured output, and diff assessment
- Configurable concurrency and invocation gap defaults (3 concurrent, 5000 ms gap)
Recommended MCP Servers
Journey fit
Canonical shelf is Build because you wire stdio MCP into the coding agent and depend on a local Gemini CLI binary—that is agent-tooling setup, not a ship-only gate. It is an MCP bridge to Gemini CLI subprocesses (concurrency and rate-gap controls), which fits agent-tooling rather than a single SaaS API integration.
How it compares
MCP bridge to Gemini CLI, not a hosted multi-model API skill baked into one vendor’s chat.
Common Questions / FAQ
Who is gemini-mcp-bridge for?
Solo and small-team builders who run MCP-capable agents and want Gemini CLI as callable tools for review, search, and structured answers.
When should I use gemini-mcp-bridge?
Use it during implementation and ship prep when you want a second model to review diffs, search the tree, or return JSON without leaving your agent.
How do I add gemini-mcp-bridge to my agent?
Install the npm package gemini-mcp-bridge, ensure gemini is on PATH or set GEMINI_CLI_PATH, add the stdio MCP entry in your host config, then restart the agent.