bestdeejay-design/agent-skills / skills/prd-to-app
prd-to-app
VerifiedBuild a complete, production-ready full-stack web application from PRD documents, prototype images, and resource files. Handles the entire pipeline: system design, database schema, seed data, backend API, frontend UI, visual verification against prototypes, and deployment script generation. Use this skill whenever the user: - Provides a PRD (product requirement document) and wants a working app built - Says "build from PRD", "implement this product", "develop this app from requirements" - Has prototype images + requirements and wants full-stack implementation - Wants to turn product specifications into a running web application Triggers: 'prd to app', 'из prd в приложение', 'разработай по prd', 'сделай приложение по макетам', 'build from prd', 'implement from prd'.
PRD-to-App: Full-Stack Application Builder
Language: Respond in the same language the user uses. Code comments should match.
Build a complete, deployed web application from PRD + prototypes + resources.
The result must be fully reproducible via a single bash start.sh.
Origin: Adapted from zai-org/GLM-skills/glmv-prd-to-app (Apache 2.0).
ZHIPU API dependencies removed; replaced with Playwright + look_at for prototype analysis.
Phase 0: Material Discovery & Analysis
Before anything else, understand what you're working with.
0a. Locate all inputs
./prd.md ← Product requirement document
./prototypes/*.jpg|*.png ← UI prototype images (the visual truth)
./resources/**/* ← Images, videos, icons, and other assets
If the materials are in a different location, adapt accordingly. Read the PRD fully.
0b. Deep prototype analysis
For every prototype image:
-
Read the image using the
look_attool (orReadtool for images) — examine it directly. Extract: page identity, layout structure, component inventory, content inventory, color extraction (hex values), typography, interactive states, data patterns. -
For each image, document:
- Page identity: which page/view this represents
- Layout structure: header, sidebar, main content, footer, modals
- Component inventory: every button, form, card, table, list, nav element
- Content inventory: all visible text, numbers, labels, placeholder content
- Color extraction: primary, secondary, accent, background, text colors (hex values)
- Typography: font sizes, weights, hierarchy observed
- Interactive states: hover effects, active tabs, selected items, toggles
- Data patterns: what data populates lists/tables/cards — this drives seed data
-
Build a page map showing navigation flow between prototype pages.
0c. Resource inventory
List all files in ./resources/ and map each to where it appears in the prototypes.
Every resource file must be used in the final application where relevant.
Phase 1: System Design Document
Produce a comprehensive design document at ./docs/design.md.
1a. Data Model
For each entity, specify:
- Table/collection name
- All fields with types, constraints, defaults
- Relationships (foreign keys, many-to-many)
- Indexes needed for query patterns
- Content mapping: which prototype elements map to which fields
1b. API Design
For every page interaction, define an API endpoint:
- Method + path
- Request params/body schema
- Response schema with example
- Which prototype interaction triggers this API
- Error responses
1c. Frontend Architecture
- Component hierarchy (tree structure)
- Route definitions mapping to prototype pages
- State management approach
- How each prototype page maps to components
1d. Technology Stack
Choose based on PRD complexity:
| Layer | Choice | When to use |
|---|---|---|
| Frontend | React + TypeScript + Vite | Default for SPAs |
| Frontend | Next.js | If SSR/SEO needed |
| Styling | Tailwind CSS | Default |
| Backend | Node.js + Express | Simple APIs |
| Backend | Python + FastAPI | If PRD mentions Python |
| Database | SQLite | Simple apps, <10 tables |
| Database | PostgreSQL | Complex apps, relationships |
| ORM | Prisma (Node) / SQLAlchemy (Python) | Match backend |
Phase 2: Seed Data Generation
Rules
- Extract from prototypes: Every piece of visible text, image, number in the prototype images must appear in seed data. Re-read each prototype image and transcribe content.
- Complete coverage: Every list/table/card/dropdown must match prototype content exactly.
- Use resource files: Map resource files from
./resources/to seed data entries. - No placeholders: No "Lorem ipsum", no "Test Item 1", no placeholder images.
- Support all states: Include data for empty states, loaded states, error scenarios.
Phase 3: Backend Implementation
3a. Database schema — migrations with constraints, indexes, foreign keys
3b. API endpoints — route handlers, validation, error handling, curl-tested
3c. Seed data loading — idempotent re-seeding, dependency order
3d. Static file serving — backend serves resource files
Phase 4: Frontend Implementation
4a. Global styles and tokens — color variables, typography, spacing from prototypes
4b. Page-by-page implementation
For each prototype image:
- Re-read the prototype image
- Build page component matching layout exactly
- Wire up API calls
- Implement all interactions (navigation, forms, search, filter, sort, modals, states)
4c. Resource integration — copy from ./resources/, reference correctly
4d. Responsive — match prototype viewport, breakpoints if mobile views shown
Phase 5: Visual Verification Loop
Repeat for every page. Max 3 iterations.
5a. Render page to screenshot
Start local server, capture with Playwright:
python3 -c "
from playwright.sync_api import sync_playwright
with sync_playwright() as p:
browser = p.chromium.launch()
page = browser.new_page(viewport={'width': 1280, 'height': 800})
page.goto('http://localhost:3000/page-path')
page.wait_for_load_state('networkidle')
page.screenshot(path='docs/screenshots/page_name.png', full_page=True)
browser.close()
"
5b. Visual comparison
Read prototype image and screenshot side by side. Compare:
- Layout, colors, typography, content, spacing, images, components
5c. Fix discrepancies → re-render → re-compare
Phase 6: Integration Testing
6a. API health check — curl each endpoint, verify status codes
6b. E2E flow — walk through every user flow from PRD
6c. Fix issues — CORS, URL mismatches, data format problems
Phase 7: Deployment Script
Generate ./start.sh — fully self-contained, works from absolute zero.
Requirements: install deps → setup DB → migrations → seed → build frontend → start both → app at :3000
Phase 8: Documentation
./docs/design.md— final architecture, data model, API reference./README.md— overview, stack, quick start, structure
Deliverables Checklist
- All PRD features implemented
- All prototype pages built
- Visual match verified via screenshots
- All resources used
- Seed data matches prototypes
- All API endpoints working
- All interactions functional
-
bash start.shworks from clean state - Documentation complete
Critical Principles
- Prototypes are truth — prototype wins over PRD text for visual/layout decisions
- No shortcuts on data — all visible content from database via APIs
- Complete implementation — every page, feature, interaction
- Resources must be used — matching files from
./resources/ - Reproducibility —
start.shworks from absolute zero - Verify, don't assume — screenshot comparison + API checks + startup test
Boundaries
- Do not use for a small isolated UI change or documentation-only task.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |