Task Breakdown

Language Configuration

Before generating any content, check aico.json in project root for language field to determine the output language. If not set, default to English.

Process

1. Read story/PRD: Load from docs/reference/pm/stories/ or docs/reference/pm/versions/
2. Identify data entities: What domain objects are needed
3. Identify API endpoints: What routes are needed
4. Identify business logic: What services are needed
5. Break into tasks: Follow layered architecture order
6. Save output: ALWAYS write to docs/reference/backend/tasks/{story-name}.md

Layered Architecture Order

1. Data Models (types, entities, DTOs)
      ↓
2. Database (migrations, schema)
      ↓
3. Repository Layer (data access)
      ↓
4. Service Layer (business logic)
      ↓
5. API Layer (controllers, routes)
      ↓
6. Validation & Error Handling
      ↓
7. Tests (unit, integration, API)

Task File Template

# [Story Name] - Backend Tasks

> Project: [project-name]
> Created: YYYY-MM-DD
> Last Updated: YYYY-MM-DD
> Source: docs/reference/pm/stories/[story].md
> Status: in_progress

## Progress

| #   | Task                       | Status     | Notes |
| --- | -------------------------- | ---------- | ----- |
| 1   | Define data models         | ⏳ pending |       |
| 2   | Create database migrations | ⏳ pending |       |
| 3   | Implement repository layer | ⏳ pending |       |

## Tasks

### Task 1: Define data models

**Status**: ⏳ pending
**Goal**: Create types for domain entities
**Scope**: Entity types, DTOs, validation schemas
**Acceptance Criteria**:

- [ ] Types match business requirements
- [ ] No type errors
      **Dependencies**: None

Task Types

Granularity Rules

• Each task = independently testable
• Each task = single responsibility
• Each task = clear scope (not too big, not too small)

Key Rules

• ALWAYS follow layered architecture order
• MUST include test tasks for each layer
• ALWAYS note dependencies between tasks
• MUST save to docs/reference/backend/tasks/ directory

Common Mistakes

• ❌ Tasks too large (entire API) → ✅ Break into layers
• ❌ Skip data model first → ✅ Types before implementation
• ❌ Skip repository layer → ✅ Separate data access
• ❌ Forget validation → ✅ Always include validation task