CONTRIBUTION.md

Contribution

Quick Start

1. Clone the repo
2. Copy .env.example to .env.local and add your API keys
3. Install dependencies: pnpm install
4. Install Python CAD engine: pip3 install build123d
5. Run: pnpm dev

Architecture

Read docs/architecture.md for the full technical architecture.

The key principle: agents are specialized and observable. Each agent has one job, logs its work, and passes structured data to the next.

Code Style

• TypeScript strict mode
• Tailwind CSS for styling (no CSS modules)
• Server components by default, 'use client' only when needed
• API routes in app/api/ — each is a self-contained module
• Zustand for client state — single store in lib/store.ts

Adding a New Agent

1. Add the model config to lib/ai-clients.ts
2. Add the agent step to app/api/generate/route.ts (emit SSE events)
3. Add the agent card to components/parametric/agent-monitor.tsx
4. Update types in lib/types.ts if needed

Adding a New Primitive

1. Add the op to the DesignNode.op union type in lib/types.ts
2. Add it to the Tree Builder's allowed ops list in app/api/generate/route.ts
3. Add Build123d example code to the Code Generator's system prompt
4. Update scoring heuristics in lib/scoring.ts if needed

Adding a New DIR Family

1. Add the family to the DIR schema in app/api/analyze-image/route.ts
2. Add the family name mapping in dirToPrompt()
3. Add the family to the VLM prompt's enum list
4. Document it in docs/dir-pipeline.md