GTAI-1/flow-mcp

Drive Google Flow from Claude, Codex or a local control panel, on Mac or Windows — using your Google AI subscription credits instead of the pay-per-use video API. No browser extension.

What it solves

This tool allows you to control Google Flow—a web-based video generation platform—directly from AI assistants like Claude or Codex. It bridges the gap between your chat interface and the Flow website, enabling you to generate, edit, and manage video projects using your own Google account credits without needing to use paid video APIs.

How it works

It runs a small program on your local machine that launches a dedicated, private Chrome window. By using the Chrome DevTools Protocol, it interacts with the actual buttons and menus on the Flow website as a human would. It connects to your AI assistant via the Model Context Protocol (MCP), allowing you to issue commands in plain English to create characters, generate shots, narrate, and edit videos. It also includes a local control panel for manual management and handles tasks like captioning and video assembly entirely on your computer.

Who it’s for

It is for users who want to automate their video production workflow in Google Flow using AI assistants. It is ideal for those who prefer to keep their media processing local and want to leverage their existing daily Google account credits rather than paying for external video generation services.

Highlights

  • Uses your own Google account credits (50 free credits daily).
  • No browser extensions or hosted services required; everything runs locally.
  • Supports complex workflows like character consistency, shot chaining, and video-to-video editing.
  • Includes local tools for narration, captioning (via whisper.cpp), and video assembly.
  • Provides a visual control panel for manual project management alongside AI-driven automation.
  • Compatible with any MCP-enabled assistant, including Claude Desktop and Codex.

Related

  • Project
    VAST-AI-Research/EdenLive

    EdenLive is an open‑source system that turns low‑fidelity reference videos (e.g., a game’s proxy rendering) into high‑quality, real‑time generative video. It provides 5 B and 14 B diffusion models, supports both bidirectional and chunk‑autoregressive inference, and includes a streaming WebRTC service, a text‑to‑game skill, and integration with Reactor and LightX2V for accelerated deployment.

  • Project
    elder-plinius/GL4SS

    A spatiotemporal image and video engine that generates AI visuals of any location on Earth at any point in time from 252 million years ago to 3050 AD.

  • Project
    seydx/node-av

    NodeAV is a native Node.js binding for FFmpeg with full TypeScript support, providing type-safe options for codecs, formats, filters, and bitstream filters. It offers both low-level and high-level APIs, hardware acceleration, device capture, and prebuilt binaries for Windows, Linux, and macOS.

  • Project
    jonghwanhyeon/python-ffmpeg

    python-ffmpeg is a Python binding that provides synchronous and asynchronous APIs to control FFmpeg, letting developers build and run media processing commands (like transcoding or recording) with a clean, chainable interface and support for progress events.