flowgen

FlowGen Image Studio

FlowGen Emblem

FlowGen Image Studio

English · Architecture · Studio GUI · Agent MCP · Quickstart · Credits

Turn your Google Flow account into an unlimited, high-speed batch image generation pipeline.
Parallel Chromium worker tabs, zero-overwrite sequential naming, 14 aesthetic styles, a modern zero-scroll cockpit,
and a native Model Context Protocol (MCP) server for AI coding agents.

Version License Python Engine MCP Protocol Google Colab Author


FlowGen Studio Interface Cockpit


💡 Why FlowGen?

Commercial image generation APIs charge steep monthly subscriptions and per-image metering ($0.04 to $0.12 per generation), while consumer browser interfaces force you to manually type, wait, copy, and paste one image at a time.

FlowGen bridges this gap completely for free:

  1. Connects your existing browser account via Chrome DevTools Protocol (CDP).
  2. Spawns parallel Chromium worker tabs to run multiple prompts concurrently (2x to 4x speedup).
  3. Automates sequential downloading with zero image overwrites.
  4. Serves as a native Model Context Protocol (MCP) endpoint so agents like Cursor, Claude Desktop, and Antigravity can generate assets autonomously.
  5. Runs anywhere: on your local PC, on a cloud VPS via Docker, or 100% free on Google Colab with Cloudflare tunnels.

⚡ How it works

Paste your prompt list, select an aesthetic style, and press Start Generating. FlowGen processes the queue across four discrete phases:

  1. Matrix Ingestion: Parses raw prompt lines, removes duplicates, and injects selected visual style tokens (e.g. 35mm film, cyberpunk neon, anime studio) with custom prefixes/suffixes.
  2. Parallel Tab Dispatch: Distributes generation tasks evenly across 1 to 4 isolated Chromium worker tabs via Chrome DevTools Protocol.
  3. Sequential Slugging: Captures rendered outputs, generates clean sequential file names (e.g. 1. Prompt_Title.png), and isolates them into dedicated project workspaces.
  4. Library & Telemetry: Updates the live canvas, logs speed and ETA in real time, and exposes generated images immediately in the Generated Library and to connected AI agents.
FlowGen Studio Generation Cockpit
Generate Cockpit: Dual collapsible sidebars, 14 aesthetic styles, and live telemetry.
Generated Library by Project
Generated Library: Filter by project folder, inspect metadata, and open in Explorer.

🎛️ Two ways to use it

1. Studio Cockpit (Desktop GUI)

Sidebars Collapsed - Max Canvas View
Max Canvas View: Sidebars collapsed with drawer pull tabs for large-scale prompt matrices.
High Contrast Light Mode
Light Theme: Crisp high-contrast zinc/slate design system for bright environments.

2. Headless Agent Engine (MCP & REST API)


🤖 Built for agents

FlowGen is built from the ground up for AI coding assistants. By exposing its tools over the Model Context Protocol (MCP), agents in Cursor, Claude Desktop, and Antigravity can write prompts, trigger batch runs, monitor generation progress, and inspect finished image files autonomously.

Claude Desktop Configuration

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "flowgen": {
      "command": "python",
      "args": ["E:/Organized/flowgen/browser2api/run_mcp.py"],
      "env": {
        "FLOWGEN_API_KEY": "your_api_key_here"
      }
    }
  }
}

Cursor IDE Configuration

In Cursor Settings → Features → MCP Servers → Add New MCP Server:

Exposed MCP Tools

| Tool Name | Parameters | Description | | :— | :— | :— | | flow_generate | prompts, model, orientation, workers, style_suffix | Submits batch generation and returns local file paths | | flow_get_status | none | Returns engine health, worker count, and active queue state | | flow_get_history | none | Returns historical analytics, image count, and daily trends | | flow_list_projects | none | Lists all project folders and generated image counts | | flow_list_styles | none | Returns available visual style presets | | flow_auth_status | none | Verifies active Google Flow session status |


🔧 Under the hood

FlowGen replaces cumbersome manual web automation with an asynchronous multi-tab orchestration engine:

┌─────────────────────────────────────────────────────────────┐
│                    FLOWGEN IMAGE STUDIO                     │
├──────────────────────────────┬──────────────────────────────┤
│      Desktop Web Cockpit     │     AI Agent MCP Server      │
│  (FastAPI + Tailwind + JS)   │   (Claude, Cursor, Agents)   │
└──────────────┬───────────────┴──────────────┬───────────────┘
               │                              │
               ▼                              ▼
┌─────────────────────────────────────────────────────────────┐
│              Orchestrator & Multi-Worker Pool               │
├─────────────────────────────────────────────────────────────┤
│  Worker Tab #1  │  Worker Tab #2  │  Worker Tab #3  │  ...  │
│  (CDP Session)  │  (CDP Session)  │  (CDP Session)  │       │
└──────────────┬──────────────────────────────┴───────────────┘
               │
               ▼
┌─────────────────────────────────────────────────────────────┐
│             Google Flow Image Generation Engine             │
└─────────────────────────────────────────────────────────────┘

🚀 Get started

Simply double-click START_FLOWGEN.bat (or run start.bat) from the root folder. The script automatically detects the virtual environment, starts the server on port 8787, and opens your default browser.


Option 2: Google Colab Cloud Run (Free Cloud GPU)

Run FlowGen completely free without using your local PC:

  1. Open FlowGen_Colab.ipynb in Google Colab.
  2. Run the cells to install dependencies and initialize the Cloudflare tunnel.
  3. Click your live public HTTPS link and copy the generated FLOWGEN_API_KEY.

Option 3: Manual Setup (Windows, macOS, Linux)

Requires Python 3.11 or newer.

# 1. Clone repository
git clone https://github.com/AliRash3ed/flowgen.git
cd flowgen

# 2. Create and activate a virtual environment
python -m venv .venv
# On Windows:
.venv\Scripts\activate
# On macOS / Linux:
source .venv/bin/activate

# 3. Install dependencies & Playwright Chromium
pip install -r requirements.txt
playwright install chromium

# 4. Launch Studio
python browser2api/run_hud.py

Open http://127.0.0.1:8787 in your browser.


Option 4: Docker VPS Deployment

docker compose up -d

The studio will be available at http://YOUR_SERVER_IP:8787.


🎨 Supported Models & Aspect Ratios

Generation Models

Canvas Aspect Ratios


🎭 Aesthetic Style Matrix

FlowGen includes 14 built-in aesthetic presets that can be injected at the beginning or end of your prompts:

Preset Name Preview Accent Signature Visual Injections
Cinematic Raw #8b5cf6 Violet 35mm lens, natural dramatic rim lighting, shallow depth of field, 8k
Cyberpunk Neon #06b6d4 Cyan Vibrant volumetric neon glow, wet night street reflections, octane render
Anime Studio #f59e0b Amber Studio Ghibli & Makoto Shinkai aesthetic, luminous clouds, painterly textures
Dark Fantasy #a855f7 Purple Eldritch gothic atmosphere, ominous fog, moody chiaroscuro lighting
Octane 3D #10b981 Emerald Subsurface scattering, translucent materials, chromatic aberration, caustics
Impressionist Oil #ec4899 Pink Heavy impasto palette knife strokes, rich pigment layers, canvas texture
Vector Graphic #3b82f6 Blue Clean isometric geometric outlines, vibrant flat colors, minimalist
Vintage 1970s #f97316 Orange Authentic Kodachrome film grain, warm nostalgic color grade, retro vignette
Photoreal Macro #14b8a6 Teal Extreme close-up lens, microscopic surface details, water droplet caustics
Editorial Fashion #6366f1 Indigo High-fashion editorial photography, studio strobe lighting, Vogue aesthetics
Minimalist Architecture #64748b Slate Monolithic concrete structures, sharp geometric shadows, clean Brutalism
Watercolor Wash #0ea5e9 Sky Expressive translucent pigment bleeding, wet-on-wet textured paper
3D Claymation #eab308 Yellow Handcrafted clay textures, stop-motion plasticine finish, studio lighting
Retrowave Synth #d946ef Fuchsia 1980s neon wireframe grid, purple chrome sunset, synthwave aesthetic

📁 Repository Structure

flowgen/
├── assets/                  # Studio brand logos, icons, and interface screenshots
│   ├── logo.svg             # Vector dark logo
│   ├── logo-light.svg       # Vector light logo
│   ├── logo.png             # Transparent FG emblem
│   ├── studio_cockpit.png   # Main studio cockpit screenshot
│   ├── studio_collapsed.png # Collapsed sidebars screenshot
│   ├── studio_light.png     # Light theme screenshot
│   └── gallery.png          # Generated library screenshot
├── browser2api/             # Core Python package & GUI engine
│   ├── run_hud.py           # GUI application launcher
│   ├── run_mcp.py           # Model Context Protocol server
│   └── src/browser2api/     # FastAPI server, multiworker pool & scrapers
├── colab_run.py             # Cloudflare tunnel & Colab runner
├── docker-compose.yml       # Container orchestration configuration
├── Dockerfile               # Production container image
├── FlowGen_Colab.ipynb      # 1-Click Google Colab Notebook
├── requirements.txt         # Core Python dependencies
├── START_FLOWGEN.bat        # 1-Click Windows desktop launcher
├── start.bat                # Shortcut launcher
├── CONTRIBUTING.md          # Open-source contribution guidelines
├── SECURITY.md              # Security & local session policy
├── LICENSE                  # MIT License
└── README.md                # Project documentation

👨‍💻 License and credits

Architected & engineered by Ali Bhatti (@alirash3ed).

Released under the MIT License. Free for personal, commercial, and research use.