| name | clone-spec-writer |
| description | Generate full technical specification (TZ) from architecture analysis |
Clone Spec Writer
When to Use
- Architecture analysis complete (output of clone-analysis)
- Ready to define what exactly to build
- Need a document a developer can implement from without ambiguity
Process
- Feature inventory -- extract every feature from analysis, tag as:
- MVP (must-have for launch)
- V2 (nice-to-have, post-launch)
- Skip (not worth cloning)
- API contracts -- for each endpoint define:
- Method, path, auth requirement
- Request body (JSON schema)
- Response body (JSON schema)
- Error codes and messages
- Data models -- define all entities:
- Fields, types, constraints
- Relationships (1:1, 1:N, N:M)
- Indexes and unique constraints
- UI wireframes description -- for each screen:
- Layout (text description or ASCII)
- Components used
- User interactions and state transitions
- Acceptance criteria -- per feature, write testable AC:
- Given / When / Then format
- Edge cases explicitly listed
- Tech stack recommendation -- based on analysis, recommend:
- Framework, language, database
- AI model provider and fallback
- Hosting and deployment strategy
Output
- Complete TZ document (markdown, 30-80 pages equivalent) containing:
- Feature list with priorities
- API contract catalog
- Data model diagrams
- UI descriptions per screen
- Acceptance criteria per feature
- Recommended tech stack with rationale
Example
Gamma.app TZ: 18 MVP features (AI generation, template picker, slide editor, export PDF, share link), 9 V2 features (collaboration, analytics, custom fonts). Data models: User, Workspace, Presentation, Slide, Theme, Asset. API: 22 MVP endpoints. Stack recommendation: Next.js + FastAPI + PostgreSQL + Gemini API.