flow-agent
Health Uyari
- No license — Repository has no license file
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 43 GitHub stars
Code Gecti
- Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
- Permissions — No dangerous permissions requested
Bu listing icin henuz AI raporu yok.
⚡ CLI toolkit for Google Flow — Nano Banana Pro images, Omni Flash videos, MCP v2 & OpenAI API.
⚡ Flow Agent
CLI, OpenAI-compatible API, Chrome extension, and MCP server for Google Flow image/video generation.
Features
- Text-to-image and reference-based image generation
- Text-to-video, image-to-video, reference-to-video, and video editing
- 4/6/8/10-second video generation
- OpenAI-compatible HTTP API
- CLI for full control
- MCP v2 with status, credits, models, history, generation, upload, URL download, and video editing
- Chrome extension bridge using your logged-in Google Flow session
gem_pix_2(Pro) as the default image model
Install from source
Requirements: Python 3.10+, Chrome, Google Flow access, and the uv package manager.
git clone https://github.com/kodelyx/flow-agent.git
cd flow-agent/flow-agent
cp config.env.example config.env
uv tool install --force .
Start the backend:
flow serve
Chrome extension
- Open
chrome://extensions. - Enable Developer mode.
- Click Load unpacked.
- Select the
flow-extensionfolder. - Open https://labs.google/fx/tools/flow, sign in, and keep the page open.
- Verify with
flow status.
Extension-specific instructions: flow-extension/README.md.
CLI
flow status
flow credits
flow image "a cinematic neon city" --model gem_pix_2 --aspect landscape
flow image "restyle this character" --ref character.png --count 2
flow video "a dragon flying over mountains" --aspect landscape --duration 8
flow video "the character starts walking" --start character.png
flow video "transition between scenes" --start first.png --end last.png
flow upload clip.mp4
flow edit "transform into a dark anime style" --media-id MEDIA_ID --video-file clip.mp4
Run flow <command> --help for every option.
MCP v2
{
"mcpServers": {
"flow": {
"command": "flow-mcp",
"args": []
}
}
}
Available tools:
get_flow_statusget_flow_creditslist_flow_modelsget_flow_historygenerate_flow_imagegenerate_flow_videoupload_flow_mediadownload_media_from_urledit_flow_video
download_media_from_url accepts direct or signed image/video links, saves them locally, and can optionally upload them to Flow.
More client configurations: MCP.md.
HTTP API
Default base URL: http://127.0.0.1:8001
| Endpoint | Purpose |
|---|---|
GET /health |
Backend and extension health |
GET /v1/models |
Available models |
GET /v1/credits |
Connected-account credits |
GET /v1/history |
Generated media history |
POST /v1/images/generations |
Generate images |
POST /v1/videos/generations |
Generate videos |
POST /v1/upload |
Upload image/video references |
Default model
IMAGE_MODEL=gem_pix_2
Available aliases:
harbor_seal/litenarwhal/standardgem_pix_2/pro
Project layout
flow-agent/
├── flow-agent/ # CLI, API, MCP and Flow generation backend
├── flow-extension/ # Chrome extension bridge
├── .github/workflows/ # Cross-platform release builds
├── MCP.md # MCP client setup
└── README.md
Security
- Never commit browser cookies, GitHub tokens, tunnel credentials, or generated private media.
- The repository ignores local
config.env,github-token,cloudflared/, outputs, logs, and build artifacts. - Set
SERVER_API_KEYbefore exposing the API outside localhost.
License
Use Google Flow and generated media according to Google's applicable terms.
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi