crisng95/flowkit

AI agent can create video content better than you. No reason why you do it manually in google flow!

What it solves

Flow Kit is a system designed to automate the generation of high-quality AI videos using Google Flow. It solves the difficulty of programmatically accessing Google Flow's internal APIs, which require session cookies and reCAPTCHA tokens that cannot be replayed outside a browser. It also provides a structured pipeline to maintain visual consistency (characters and locations) across multiple scenes in a long-form video project.

How it works

The system uses a FastAPI-based Python agent that communicates via WebSockets with a custom Chrome extension. The extension acts as a browser bridge, executing RPC calls (batchexecute) directly within a signed-in flow.google.com tab.

Beyond the browser bridge, it implements a full production pipeline:

  1. Story & Entities: Defines the narrative and key characters/props.
  2. Reference Images: Generates a single image for each entity to ensure visual consistency.
  3. Scene Generation: Creates images based on those references.
  4. Video Clips: Converts scene images into 8-second video clips with motion and sound.
  5. Post-Processing: Handles 4K upscaling, TTS narration, and concatenation into a final video with YouTube-optimized thumbnails and metadata.

Who it’s for

It is built for AI video creators and developers who want to automate the production of narrative-driven AI videos with consistent characters, as well as users of AI coding agents (like Claude Code or Muse) who can use the provided Markdown "skills" to drive the pipeline.

Highlights

  • Browser Bridge: Bypasses API restrictions by running requests inside a live Chrome tab.
  • Visual Consistency: Uses a reference image system to keep characters and faces identical across different scenes.
  • AI Agent Integration: Includes 36 Markdown-based skills that allow AI coding agents to orchestrate the entire video production process.
  • Multi-Provider Support: Can route generation through Google Flow, a separate muse2api gateway, or a custom assistant worker.
  • Ops Console: A local React dashboard for monitoring pipeline progress, KPIs, and scene-level reviews.

Related

  • Project
    crisng95/flowboard

    Flowboard is a local‑only, graph‑based UI for building e‑commerce media pipelines that generate images and videos via Google Flow (Veo 3.1 i2v). It uses a Chrome extension to proxy signed Flow requests and a locally installed LLM CLI (Claude, Gemini, or Codex) for vision and auto‑prompting. All board data is stored in SQLite; no cloud backend is required.

  • Project
    liangdabiao/Seedance2-Storyboard-Generator

    An AI video production workflow that transforms stories into multi-episode series by coordinating script generation, consistent asset creation, and Seedance 2.0 video extension.

  • Project
    daydreamlive/scope

    A tool for running and customizing real-time, interactive generative AI video pipelines using autoregressive diffusion models and a composable architecture.

  • Project
    xuanyustudio/LocalMiniDrama

    LocalMiniDrama is a local-first AI short drama creation tool that takes a story synopsis through script, characters, scenes, storyboards, images, and video generation, finally composing full episodes — all on your own machine with your own AI API keys and no cloud upload.

  • Project
    Bomx/super-video-maker-skill

    An end-to-end AI video production skill for agentic coding frameworks that enables AI agents to create polished videos using avatars, b-roll, and automated composition.