Integrates with Google Flow to generate production-quality images and videos using Imagen 3/4 and Veo models, leveraging Google AI Pro subscription quotas. Provides tools for image generation, video generation, and checking whether the Google session is authenticated and ready.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
An MCP (Model Context Protocol) server that empowers AI agents (Antigravity, Claude Desktop, Cursor) to generate production-quality visual media using Google Flow (powered by Imagen and Veo models) completely in the background, with zero per-token API costs by leveraging your Google AI Pro subscription.
Zero API Cost: Uses your active Google AI Pro / Google Flow subscription quotas instead of metered cloud API calls.
100% Invisible Background Execution: Runs via headless browser automation or attaches seamlessly to Chrome via CDP (port 9222). No focus stealing.
V3 Studio Staging Pipeline: 3-step interactive draft review and finalization workflow (flow_finalize_asset) with automatic 2K upscaling.
Veo & Imagen Models: High-resolution image generation and cinematic video generation.
Aspect Ratio Selection: Support for 16:9, 9:16, 1:1, 4:3, and 3:4 (portrait/poster).
Batch Variation Multiplier: Request up to 4 concurrent variations per prompt (batch_count=1..4).
Veo Audio Control: Toggle native audio generation or mute soundtracks for custom audio scoring.
Cinematic Camera Director: Controls camera motion (pan_left, pan_right, tilt_up, tilt_down, zoom_in, zoom_out, orbit, static).
Prompt Expander: Automatic enrichment of prompts with professional lighting, optics, and atmospheric descriptors.
Frames-to-Video Animation: Interpolate between starting (start_frame) and final (end_frame) images.
Video Extension: Extend generated videos by another 5–8 seconds.
2K Upscaling: Request 2K versions directly or upscale previously generated images.
Project Management: List and create isolated projects directly from MCP tools.
Image-to-Image / Reference Input: Attach local reference images to guide generation or seed video clips.
Direct Asset Delivery: Automatically downloads rendered media files directly into your project's assets folder and provides Markdown embed links.
Related MCP server: google-flow-mcp
flowchart LR
Step1["1. Draft
flow_generate_image/video
(quick preview)"] --> Step2["2. Review & Edit
Chat preview embed
flow_edit_image"]
Step2 --> Step3["3. Finalize
flow_finalize_asset
(2K upscale & code export)"]git clone https://github.com/voronin-s-dev/google-flow-mcp.git
cd google-flow-mcp
uv venv
uv pip install -r requirements.txt
playwright install chromiumRun the interactive login script once to save your session cookies:
uv run python auth.pyA browser window will open. Sign in to your Google Account with Google AI Pro access, navigate to flow.google.com, and the session state will automatically be saved to storage_state.json.
Tool | Parameters | Description |
|
| Generates images via Imagen 3/4. Supports reference input, aspect ratio (including 3:4), batch multipliers (x1–x4), prompt expansion, and 2K upscaling. |
|
| Generates cinematic video clips via Veo with camera controls and optional audio muting. |
|
| Generates a video interpolating between starting and ending frames. |
|
| Extends an existing video tile by an additional 5–8 seconds. |
|
| Atomically upscales the chosen asset to 2K and saves it directly into the user's project folder. |
|
| Upscales a previously generated image in the project to 2K resolution. |
|
| Applies modifications or creates variations of an existing image. |
|
| Triggers re-generation of a tile with the same prompt and parameters. |
| none | Lists existing Google Flow projects with their IDs and direct links. |
|
| Creates a new isolated project in Google Flow. |
| none | Checks whether the session is authenticated and ready to generate. |
Local-First & Off-Cloud Execution: The MCP server runs exclusively on your local workstation. It requires no intermediary cloud proxy or external telemetry service.
Controlled Network Endpoints: All network communications are strictly bounded to:
https://flow.google.com (Google's official web studio endpoint).
127.0.0.1:9222 (Local Chrome DevTools Protocol port, bound only to the local loopback interface).
Safe Credential Isolation: Google session cookies (storage_state.json) and temporary Chrome profile data are stored locally and explicitly protected via .gitignore to prevent leakage.
MCP Tool Annotations: Every exposed tool declares strict safety hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) following MCP specifications, enabling client hosts to reason about side effects before execution.
Privacy Policy: See detailed privacy terms in PRIVACY.md.
Run the test suite and contract harness:
# Run unit & contract tests
uv run pytest
# Run contract and live CDP selector verification
uv run python scripts/verify_harness.pyMIT License. Feel free to use, modify, and distribute.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Build and run visual creative-production workflows from your AI agent.
AI image + video generation for agents: --flag prompt DSL, async generate/poll, x402 pay-per-use.
Generate and edit images, video, voice, lip-sync and 3D models from your AI agent.
On-brand creative studio for AI agents: images, video, audio, and 3D.
