Human Docs Generator Skill
Role
Automatically generates and maintains role-specific documentation for humans when significant changes occur in the system.
Purpose
Transform complex AI state and technical implementations into simple, role-specific guides that humans can actually understand and use.
Trigger Conditions
Automatic Triggers
- •✅ Epic completed
- •🔧 Public API modified
- •🎯 New user feature deployed
- •💥 Breaking change introduced
- •🐛 Critical bug fixed
- •📊 5+ related tasks completed
Manual Trigger
json
{
"event": "generate.human.docs",
"scope": "all|specific-role",
"reason": "manual request"
}
Document Structure
Files Maintained
code
ai-state/human-docs/ ├── INDEX.md # Role directory & quick nav ├── frontend-developer.md # React, components, routes ├── backend-developer.md # APIs, models, database ├── architect.md # System design, patterns ├── devops.md # Deploy, monitor, debug ├── tester.md # Test suites, coverage ├── product-manager.md # Features, metrics, roadmap └── end-user-guide.md # Screenshots, how-to-use
Generation Process
1. Gather Context
yaml
sources: - ai-state/active/ # Current implementations - ai-state/knowledge/ # Patterns and decisions - operations.log # Recent changes - test-results/ # What's working
2. Extract Role-Relevant Info
Frontend Developer Needs
- •Component locations & hierarchy
- •State management patterns
- •Reusable components list
- •Recent UI changes
Backend Developer Needs
- •API endpoint documentation
- •Database schema changes
- •Authentication flow
- •Service dependencies
End User Needs
- •Screenshots with annotations
- •Step-by-step guides
- •Feature locations
- •Common workflows
3. Transform to Human Format
Rules for Human Docs
- •Max 1000 words per document
- •Diagrams over text (Mermaid.js)
- •Examples over explanations
- •Current state only (no history)
- •Quickstart focused
4. Generate Diagrams
Component Tree (Frontend)
mermaid
graph TD
App --> Header
App --> Routes
Routes --> Login
Routes --> Dashboard
Dashboard --> Charts
Dashboard --> Tables
API Flow (Backend)
mermaid
sequenceDiagram
User->>Frontend: Click Login
Frontend->>API: POST /auth/login
API->>Database: Validate credentials
Database->>API: User data
API->>Frontend: JWT token
Frontend->>User: Dashboard
User Flow (End User)
mermaid
graph LR
Start[Open App] --> Login[Enter Email/Pass]
Login --> Dash[View Dashboard]
Dash --> Upload[Upload Data]
Upload --> Process[Wait for Processing]
Process --> View[View Results]
View --> Export[Download Report]
Document Templates
Frontend Developer Template
markdown
# Frontend Developer Guide ## Quick Start [3-step setup process] ## Component Map [Mermaid diagram of components] ## To Add a New Page 1. Create component in `/src/pages/` 2. Add route in `/src/routes.tsx` 3. Update navigation in `/src/components/Nav.tsx` ## Available Components - `<Button>` - Primary actions - `<Card>` - Content containers - `<Table>` - Data display [List reusable components] ## State Management [Current state shape diagram] ## Recent Changes - ✅ Added password reset - ✅ Updated dashboard layout
End User Template
markdown
# How to Use [App Name] ## Getting Started [Screenshot of login page with arrows] ## Main Features ### Upload Data 1. Click "Upload" button (top-right) 2. Select your CSV file 3. Click "Process" [Screenshot with numbered steps] ### View Reports [Screenshot of reports page] - Yellow box = Warnings - Green box = Success - Red box = Errors ## Common Tasks **Q: How do I reset my password?** A: Click "Forgot Password" on login page **Q: Where are my old reports?** A: Click "History" in sidebar
Quality Checks
Before Publishing
- • Under 1000 words?
- • Has diagrams/screenshots?
- • Includes quickstart?
- • Examples provided?
- • No technical jargon?
- • Tested instructions work?
Integration
Input Events
json
{
"event": "epic.completed",
"epic_id": "epic-oauth",
"tasks_completed": ["task-001", "task-002"],
"changes": ["Added Google login"]
}
Output Events
json
{
"event": "human.docs.updated",
"files": [
"frontend-developer.md",
"end-user-guide.md"
],
"sections_changed": [
"authentication",
"login-methods"
]
}
Anti-Patterns to Avoid
❌ Writing changelogs (just current state) ❌ Technical deep-dives (keep it simple) ❌ Walls of text (use bullets/diagrams) ❌ Assuming knowledge (explain everything) ❌ Version history (one living document)
Success Metrics
- •Reading time < 5 minutes
- •Diagram-to-text ratio > 40%
- •Quickstart success rate > 90%
- •Human confusion tickets < 5%
- •Doc update frequency = feature frequency