gflow-cli
A command-line client and MCP server for Google Flow: Veo video and Imagen images from the terminal, in batches or from agents
- 347 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The utility drives a browser holding your Google session, and automation falls under Google's terms
- Video generation spends account credits, and Flow interface errors can interrupt a batch midway
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/gflow-cliDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The utility drives a browser holding your Google session, and automation falls under Google's terms
- Video generation spends account credits, and Flow interface errors can interrupt a batch midway
This is third-party code. Review the repository files before installing.
What it does
gflow-cli drives Google Flow without manual clicks. The Python utility opens a real browser with your Flow session and starts generation: Veo makes the video and Imagen makes the images. `gflow image t2i` creates an image from text. The gflow video subcommands produce video from text, from a frame and from references, and `gflow video extend` continues a finished clip past the 8-second limit. For recurring characters there is `gflow character create`: it creates a hero in Flow with a face and body reference so a person looks the same from shot to shot. The `--tool creative-director` flag rewrites a short prompt into a rich one using a five-part formula before generation. Batch runs rely on a plain shell loop over a prompt file. Results land in the folder from GFLOW_CLI_OUTPUT_DIR or go to cloud storage through GFLOW_CLI_STORAGE_URI. The built-in MCP server gives the same operations to agents over stdio with `gflow mcp run` or over HTTP through `gflow serve`. For Claude Code and Codex there is a plugin with the gflow-cli and video-production skills.
Who it is for. For people with a Google account that has Flow access and Veo credits who run video and images in batches or hand generation to an agent.
Good fit when
- You need a series of images from a ready prompt file
- A Flow clip is shorter than the scene needs
- The same person must recur across several videos
- An agent in Claude Code or Cursor needs to start generation and fetch the result through the MCP server
- Finished files should go straight into S3, MinIO or Google Cloud Storage
Not a fit when
- You have no Google account with Flow access: the utility works only on top of your session
- You need a server-side run without a browser window: generation currently goes through a visible Chromium window
- You do not want to automate your Google account: Google's terms restrict such automation
- You need production stability: the Flow interface and endpoints may change without notice
Example request
Generate five vertical images from the prompts in prompts.txt, then make an 8-second video from the best one and show the remaining credits.Limitations
- The project is unofficial, built by reverse engineering and not affiliated with Google. It is at the alpha stage.
- It needs a Google account with Flow access, and every video generation spends that account's credits. Images and composition operations are free.
- The utility drives a visible Chromium browser through Playwright, with about 150 MB downloaded on first run. The first call takes 30 to 90 seconds.
- When the Flow interface changes, the command stops with a distinct exit code, for example 23 for selector drift, and does not resubmit the job.
- Updates install with `gflow update`.
How to disable. Remove the utility with `uv tool uninstall gflow-cli` and uninstall the plugin in Claude Code. The Flow session is stored in a Chromium profile on your machine; delete that profile if you want to sign out.
FAQ
How do I install gflow-cli?
The easiest way is `uv tool install gflow-cli`, and `pip install gflow-cli` also works. Then run `uv tool run --from gflow-cli playwright install chromium` once and sign in with `gflow auth login --browser chrome`.
How do I connect gflow-cli to Claude Code?
Run `/plugin marketplace add ffroliva/gflow-cli` and `/plugin install gflow@gflow-cli`. The plugin brings the skills and the MCP server but installs disabled. For Codex the commands are `codex plugin marketplace add .` and `codex plugin add gflow@gflow-cli`.
What costs money and what is free?
Flow credits are spent only on Veo video. Images and composition operations are free. The balance is shown by `gflow credits user`, which spends nothing.
How do I stop an agent from spending credits?
Register the server with `gflow mcp run --no-spend`. In this mode the tools that spend credits are not registered at all.
How is gflow-cli different from an official Veo API?
There is no official API here: the utility drives the Flow interface through your browser and Flow's private API. According to the developer, a pure HTTP transport for video does not work yet because of a reCAPTCHA check and sign-in through Chrome.
Discussion
No questions or reviews yet.
Related
CLI toolsAn MCP server through which an agent makes frames and videos in Google Flow under your Google account
Skills, a CLI and an MCP server that let Claude Code or Codex make video from a brief: editing, graphics, voiceover, subtitles
ClipForge MCP
ClipForge
ClipForge MCP builds a vertical video from a topic or a product link: stock footage, Edge TTS voiceover and captions, no keys, composed with FFmpeg
Two skills for generative video and images: video writes Seedance, Kling and Veo prompts by dramaturgy rules, image picks the model and its syntax
A set of skills for agents to generate and edit images, video and audio through the muapi.ai service and muapi-cli
A skill that writes prompts for video, image and music generation on 14 platforms and can submit them in the browser
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| gflow-cli (this entry) | no | 347 | ||
| MiniMax MCP | yes | API key | 1,594 | the official MiniMax MCP: speech, voices, image and video generation |
| ima2-gen | no | 889 | a local image and video generation studio for GPT, Grok, Gemini, NovelAI and ComfyUI | |
| Nano Banana 2 Skill | no | 416 | a CLI and Claude Code skill for Gemini images from 512 to 4K |
README in short
The documentation opens with a 60-second quick start and a separate block of warnings about the account and the cost of mistakes. Several ways to connect AI agents get their own section.