| name | cook |
| description | Meta-orchestrator for feature implementation. Analyzes feature requests and auto-selects the best combination of skills from .agents/skills/ to execute. Use when building new features, adding functionality, creating UI components, implementing APIs, or any constructive development task. |
| argument-hint | Feature description, e.g. 'add comment system with real-time updates' or 'build admin analytics dashboard' |
Cook — Feature Orchestrator
Analyze user's feature request, select the optimal skill combination, then execute.
Skill Catalog
Core Implementation
| Skill | Triggers |
|---|
physthink-feature | New resource, CRUD, full-stack feature, Prisma model, oRPC procedure |
backend-development | API design, auth, DB queries, microservices, security, CI/CD |
databases | Schema design, queries, migrations, indexes, aggregation |
better-auth | Login, signup, OAuth, 2FA, sessions, RBAC |
ai-elements | AI elements components built on top of shadcn |
Frontend & UI
| Skill | Triggers |
|---|
frontend-design | Page layout, component design, responsive UI |
ui-styling | Tailwind, shadcn/ui, dark mode, theming, accessible components |
ui-ux-pro-max | Advanced UI/UX design, user research, persona development |
shadcn | Adding/composing shadcn components, component search, registry |
web-frameworks | Next.js App Router, RSC, SSR, SSG, Turborepo, monorepo |
aesthetic | Design principles, visual hierarchy, micro-interactions |
Specialized Domains
| Skill | Triggers |
|---|
threejs | 3D scenes, WebGL, animations, shaders |
payment-integration | Checkout, subscriptions, webhooks, QR payments |
shopify | Shopify apps, extensions, Liquid, Polaris |
ai-multimodal | Audio/video/image processing, Gemini API |
mcp-builder | Building MCP servers, tool integrations |
Support
| Skill | Triggers |
|---|
docs-seeker | Need latest docs, unfamiliar library/API |
sequential-thinking | Complex planning, unclear scope, multi-stage design |
code-review | Pre-merge review, quality validation |
Procedure
1. Analyze Request
- What is being built? (resource, page, API, component, integration)
- What layers are involved? (DB, backend, frontend, styling, 3rd-party)
- Any domain-specific needs? (3D, payments, auth, AI)
2. Select Skills
Match request against the catalog above. Select 1–4 skills — minimum needed. Always check:
- New data model →
physthink-feature
- New UI →
ui-ux-pro-max + ui-styling + shadcn + ai-elements (for AI components) + threejs (for 3D)
- Unfamiliar API →
docs-seeker first
- Complex scope →
sequential-thinking first
3. Execute in Order
- Planning skills first (sequential-thinking, docs-seeker)
- Backend skills next (physthink-feature, databases, backend-development, better-auth)
- Frontend skills last (web-frameworks, frontend-design, ui-styling, shadcn, aesthetic)
- Validation at end (code-review)
4. Load & Follow
For each selected skill, read its SKILL.md from .agents/skills/<name>/SKILL.md and follow its procedure. Skills may reference sub-files — load those as needed.
Example Dispatch
| Request | Skills Selected |
|---|
| "Add comment system" | physthink-feature → ui-styling + shadcn |
| "Build analytics dashboard" | sequential-thinking → physthink-feature → frontend-design + shadcn |
| "Add Google OAuth" | better-auth |
| "Create 3D physics sim" | threejs |
| "Add Stripe subscriptions" | payment-integration → physthink-feature |
| "Build admin course editor" | physthink-feature → ui-styling + shadcn + web-frameworks |