Noisett

AI Brand Asset Generator - Generate on-brand illustrations and icons using AI.

Built with Agent-First Development (AFD) principles—commands are the source of truth, validated via CLI before any UI surface is built.

Quick Start

# Install dependencies
pip install -e .

# Run CLI commands
noisett asset.generate '{"prompt": "cloud computing concept", "asset_type": "product"}'
noisett asset.types '{}'
noisett model.list '{}'
noisett job.list '{}'

# Check system health
noisett doctor

# List available commands
noisett commands

Architecture

┌─────────────────────────────────────────────────────────────────────────┐
│                           SURFACES (Thin Wrappers)                       │
│  ┌─────────────────┐  ┌─────────────────┐  ┌─────────────────────────┐  │
│  │   VS Code /     │  │   Web UI        │  │   Figma Plugin          │  │
│  │   Cursor (MCP)  │  │   (Vanilla JS)  │  │   (v2)                  │  │
│  └────────┬────────┘  └────────┬────────┘  └────────────┬────────────┘  │
│           │ MCP (stdio)        │ REST API               │ REST API      │
│           └────────────────────┼────────────────────────┘               │
│                                ▼                                        │
├─────────────────────────────────────────────────────────────────────────┤
│                        COMMAND LAYER (Source of Truth)                   │
│                    Python + FastMCP + Pydantic                           │
│  asset.generate │ asset.types │ job.status │ job.cancel │ model.list    │
├─────────────────────────────────────────────────────────────────────────┤
│                           ML INFERENCE LAYER                             │
│              PyTorch + Diffusers + HiDream + LoRAs                       │
└─────────────────────────────────────────────────────────────────────────┘

Commands

Command	Description	Mutation
asset.generate	Generate images from prompt	Yes
asset.types	List available asset types	No
job.status	Get generation job status	No
job.cancel	Cancel a running job	Yes
job.list	List user's recent jobs	No
model.list	List available models	No
model.info	Get model details + licensing	No

All commands return CommandResult with UX-enabling fields:

{
  "success": true,
  "data": {...},
  "reasoning": "Started generation of 4 product illustrations",
  "confidence": 0.95,
  "suggestions": ["Try 'premium' for marketing-grade quality"]
}

MCP Integration

Configure in .cursor/mcp.json:

{
  "mcpServers": {
    "noisett": {
      "command": "python",
      "args": ["-m", "src.server.mcp"],
      "cwd": "D:/Github/Falkicon/Noisett"
    }
  }
}

REST API

Start the API server:

uvicorn src.server.api:app --port 8000

Endpoints

Endpoint	Method	Description
/health	GET	Health check
/api/generate	POST	Generate images
/api/asset-types	GET	List asset types
/api/jobs/{job_id}	GET	Get job status
/api/jobs/{job_id}	DELETE	Cancel job
/api/jobs	GET	List user's jobs
/api/models	GET	List models
/api/models/{model_id}	GET	Get model info

Development

# Install with dev dependencies
pip install -e ".[dev]"

# Install with ML dependencies (for actual image generation)
pip install -e ".[ml]"

# Run tests (29 tests)
pytest tests/ -v

# Lint code
ruff check .

Implementation Status

Phase	Component	Status
1	Commands	✅ Done
2	MCP Server	✅ Done
3	ML Pipeline	✅ Done
4	REST API	✅ Done
5	Web UI	✅ Done
6	Deployment	🔜 Next

Documentation

Document	Description
CHANGELOG.md	Release history and changes
Strategy/00-overview.md	Architecture and AFD principles
Strategy/01-commands.md	Command definitions with Pydantic schemas
Strategy/02-mcp-server.md	FastMCP integration guide
Strategy/03-web-ui.md	Vanilla JS UI implementation
Strategy/04-deployment.md	Azure deployment guide

License

MIT