---

name: aid-impl-plan description: AID Phase 3 - Implementation Planning with consolidation-first approach. Resolves contradictions between PRD and Tech Spec, creates consolidated master document, then breaks down into actionable tasks and populates Jira. Includes sprint planning and risk assessment.

AID Implementation Plan Skill (Phase 3)

Templates & References

Templates:

• references/templates/epic-template.md
• references/templates/story-template.md
• references/templates/task-template.md
• references/templates/contradiction-log-template.md

Methodology:

• references/phase3-methodology.md
• references/content-mapping.md
• references/criteria-generator.md
• references/iron-rules.md

Key Principle: PRD Content → Epics & Stories (WHY/WHAT), Tech Spec Content → Tasks (HOW)

---

Golden Rules of Phase 3

Rule #1: NO WORD LEFT BEHIND

Every word from source documents MUST appear in Jira:

• PRD content → Epics/Stories
• Tech Spec content → Tasks
• OpenAPI endpoints → Implementation Tasks

Verification: Create traceability matrices. 100% coverage required before Phase 4.

Rule #2: SMALL TASKS, BIG DOCUMENTS

Rule #3: PROCESS IN CHUNKS, WRITE IMMEDIATELY

Wrong: Read all → Think → Write at end (loses info) Right: Read section → Write to enriched file → Next section (progressive capture)

Rule #4: VERIFY BEFORE PROCEEDING

Before Phase 4:

• Every PRD user story → Story in Jira
• Every Tech Spec component → Tasks
• Every Task has all required fields
• Coverage < 100%? Find gaps → Add items → Re-verify

---

Phase 3 Sub-Phases

CopyPhase 3a → Phase 3b → Phase 3c → Phase 3d
Consolidation  Breakdown  Criteria Gen  Jira Population

Phase 3a: Contradiction Resolution & Consolidation

Purpose: Find and resolve ALL contradictions between PRD, Tech Spec, and Research BEFORE consolidating.

Process:

1. Document Discovery & Section Mapping
2. Create Processing Order (dependency order)
3. Section-by-Section Consolidation
4. Contradiction Analysis (per section)
5. Resolution Hierarchy: Research > PRD > Tech Spec
6. Progressive Document Building

Contradiction Types:

Resolution Template:

Copy## Contradiction #[N]
**Found In:** [PRD section] vs [Tech Spec section]
**Description:** [What conflicts]
**Resolution:** [How resolved]
**Authority Used:** [Research/PRD/Tech Spec]
**Rationale:** [Why]

Phase 3a Checkpoint: User approval required before proceeding to 3b.

---

Phase 3b: Task Breakdown

Purpose: Transform consolidated spec into actionable development tasks.

Component Categories:

• Backend: API, Database, Business Logic, Integrations
• Frontend: Pages, Components, State, API Integration
• Infrastructure: Database Setup, Config, CI/CD
• Cross-Cutting: Auth, Logging, Error Handling

Task Decomposition Rules:

Estimation:

Risk Template:

Copy### Risk: [Name]
- **Category:** [Technical/Dependency/Resource/Timeline/Scope]
- **Probability:** [High/Medium/Low]
- **Impact:** [High/Medium/Low]
- **Mitigation:** [Strategy]
- **Contingency:** [Backup plan]

---

Phase 3c: Success Criteria Generation (QA Gate Setup)

Purpose: Generate testable acceptance criteria for QA sub-agent validation.

Information Boundaries:

Output: .aid/qa/{task-id}.yaml

Copyschema_version: "1.0"
task_id: "{TASK-ID}"
business_context:
  epic_goal: "{sanitized}"
  user_value: "{sanitized}"
  acceptance_criteria: []
criteria:
  must_achieve: []
  must_not: []
  not_included: []
  best_practices: []
files_to_review: []
review_history: []

Phase 3c Checkpoint: Human approval required before Jira population.

---

Phase 3d: Jira Population

Two-Step Process:

Step 1: Create Structure

1. Create all Epics (summary, priority, labels)
2. Create all Stories under Epics
3. Create all Tasks under Stories
4. Create all Subtasks under Tasks
5. Link all dependencies

Step 2: Enhance with Full Details For each issue, add:

• Full business context (PRD)
• Technical implementation (Tech Spec)
• Acceptance criteria
• Technical notes
• Test strategy
• Reference to consolidated spec

Enhancement Template:

Copy## Summary
[What this accomplishes]

## Business Context
[From PRD]

## Technical Implementation
[From Tech Spec]

## Acceptance Criteria
- [ ] [Criterion 1]
- [ ] [Criterion 2]

## Dependencies
- Blocks: [keys]
- Blocked By: [keys]

## Reference
📄 Consolidated Spec: [path]
📑 Section: [name]

Alternative Export: CSV/JSON for non-Jira users.

---

Lessons Learned

1. Stage enriched content in docs/implementation-plan/enriched-jiras/ before pushing to Jira
2. Never skip Story layer - Epic → Story → Task (not Epic → Task)
3. Spec fixes are NOT developer work - fix source documents directly
4. Content mapping is mandatory - PRD → Epic/Story, Tech Spec → Task
5. Verification is mandatory - run coverage agents before Phase 4
6. Use ADF for Jira - Atlassian Document Format for rich descriptions
7. Task naming: E{Epic#}-T{Sequential##}: {Verb} {Component}
8. Every task needs: Description, Files, Code pattern, API contract, Error handling, AC, Estimate, Spec reference

---

Output Artifacts

---

Exit Criteria (Phase 3 → Phase 4)

• Phase 3a: Consolidated doc created, user approved
• Phase 3b: Tasks < 4hr, acceptance criteria, dependencies mapped
• Phase 3c: QA criteria files generated, human approved
• Phase 3d: Jira populated, all issues have full descriptions
• Quality Gates: Sub-agent review PASSED

---

Commands

---

State Tracking

.aid/state.json:

Copy{
  "current_phase": 3,
  "sub_phase": "3b",
  "phase_3_state": {
    "3a": {"status": "complete", "user_approved": true},
    "3b": {"status": "in_progress"},
    "3c": {"status": "locked"},
    "3d": {"status": "locked"}
  }
}