A free tool lets Claude Code, Codex and Gemini CLI use your Google AI Pro plan

One subscription, several tools

As dev.to reported, a developer who uses Claude Code all day also pays for Google AI Pro, and noticed he hardly touched what the plan included. His explanation was simple: switching between tabs interrupts the work. So he wrote a small open-source program, called gemini-pro-mcp, that lets his coding assistant reach Google's tools directly.

The program works with Claude Code, with Codex and with the Gemini CLI, which is Google's own command-line assistant. You install it once, and each of those tools can then call Gemini without you leaving the window you are working in.

What MCP is, and what this server adds

MCP stands for Model Context Protocol. It is a common standard for handing an AI agent a set of tools. A server publishes functions, such as a request to ask Gemini a question, and the assistant (the client) decides when to use them. Because the standard is shared, one install can serve every compatible client on your machine.

The main Gemini functions go through Google's Antigravity CLI, a command-line program where you log in with your Google account. That is the key detail: those calls draw on your subscription rather than on a pay-per-use API.

What you can ask it to do

The server bundles several jobs, and the author describes them one by one:

• A second opinion. Claude can ask Gemini to review a function for race conditions, then compare the reply with its own view.
• Research with links. A research tool returns answers with sources, which helps when the assistant needs current information.
• Delegated coding. Gemini can edit files in a folder you name, working in a mode that accepts edits inside a sandboxed terminal. Claude then reviews the changes.
• Notebooks. Gemini Notebooks, formerly NotebookLM, lets you load PDFs or web pages and question them. It can also produce audio summaries, mind maps, quizzes and slide decks.
• Jules. You can start a session of Google's coding agent on a GitHub repository, approve its plan and message it, using the official API.
• Video, experimental. A request such as a two-minute video of a lighthouse keeper's day becomes a shot list. Each clip is generated in Google Flow and the clips are joined into one file.

The video feature works by operating Flow in a separate Chrome profile that you sign into once.

Where the bill can still appear

Not everything is covered by the plan. The author says the video and image tools that rely on the Gemini API are charged to your API key instead. Their descriptions say so, and the client asks for confirmation before it calls them.

The Flow video tools spend your monthly Flow credits, but only after a separate approval step. Browser-driven tools stay switched off until you turn them on. These guardrails matter because an agent that can spend money or control a browser needs a clear stop before it acts.

One caution from our own reporting: this is a third-party project, not an official Google or Anthropic feature. Rules about which tools may use a consumer plan can change, so check the terms for your plan and for each client before you rely on it. Keep your login details out of shared repositories, too.

Setting it up

You need a Google AI Pro or Ultra plan, a Python tool launcher called uv, and the Antigravity CLI signed in. Then you run one command, uvx gemini-pro-mcp setup. It looks for compatible assistants on your computer and registers the server in each.

A second command, uvx gemini-pro-mcp doctor, reports what is working and what is missing. After restarting Claude Code, the new tools should appear. The author suggests a first test: ask the assistant to list the Gemini models available on your plan.

The project is MIT-licensed, which is a permissive open-source license. It is published on PyPI, the Python package index, and in the official MCP Registry. The author invites bug reports and contributions, and the repository includes a guide for adding new tools.