Flow Agent by Kodelyx is an MCP server tool designed for programmatic interaction with Google Flow. Built primarily with Python 3.10 or newer, it enables automation of image and video generation tasks. The tool is distributed as Python source code and also offers pre-compiled standalone binaries for Windows, macOS, and Linux. It functions as both an MCP v2 server and an OpenAI-compatible API, providing developers and users with a way to control Google Flow's capabilities through code.

What does Flow Agent connect to?

Flow Agent connects directly to Google Flow, establishing a link through an existing, logged-in Google Flow browser session. This connection is facilitated by a dedicated Chrome extension bridge. The program operates as an MCP v2 server and provides an OpenAI-compatible API, specifically tailored for Google Flow's image and video generation services. Its operation relies on an active browser session where the Chrome extension captures the necessary session key for communication.

Core features and tools

The tool exposes several callable functions via its MCP v2 interface. These include capabilities for generation, upload, download, history, status, and credits management within Google Flow. It supports stdio as a transport mechanism for MCP clients. Additionally, Flow Agent offers OpenAI-compatible HTTP endpoints, which are particularly useful for handling pollable video jobs, allowing for asynchronous processing and retrieval of video content.

Who is it for?

Flow Agent is intended for developers and users who aim to automate or programmatically interact with Google Flow for media creation. This includes individuals seeking to control image and video generation, as well as those managing generated and uploaded content. It serves those who need to integrate Google Flow's creative services into larger automated workflows. Skip this if you are not comfortable with browser extension setup or maintaining an active browser session.

How to get started

Access is granted through an active, signed-in Google Flow browser session. The tool supports stdio as an MCP transport.

Use cases

• Automated text-to-image and text-to-video generation.
• Programmatic video editing, such as image-to-video and reference-to-video.
• Managing generated and uploaded media, including tracking history and credits.

Flow Agent by Kodelyx offers a specialized interface for automating Google Flow's image and video generation. It provides a structured way for developers to integrate Google Flow into their applications, relying on an active browser session for functionality. The tool's design as an MCP v2 server and OpenAI-compatible API makes it adaptable for various automation needs. Try automating a text-to-image generation task through its OpenAI-compatible API.