[PHASE 2] Implement Planner Agent Core Functionality #603

Problem

Planner agent is currently a stub (46 lines, only logs context). Needs core implementation for context analysis, plan generation, and checklist posting.

Acceptance Criteria

• Agent accepts structured input (PR/issue context)
• Analyzes PR/issue body, labels, and linked issues
• Generates structured plan based on type:
  • Architecture plan: Design → API contract → Data model → Implementation
  • Implementation plan: Setup → Core logic → Tests → Docs → Review
  • Task plan: Analysis → Research → Implementation → Verification
• Posts plan as markdown comment on PR/issue
• Dry-run mode works (doesn't post when DRY_RUN=true)
• Error handling in place (try-catch, clear errors)
• Tests pass (coverage ≥ 80%)
• Workflow re-enabled (remove if: false)

Implementation Plan

1. Define Module Interface

async function run(context = github.context, options = {}) {
  // options: { dryRun: false, planType: 'auto'|'architecture'|'implementation'|'task' }
  // returns: { success: bool, plan: string, commentId: number, errors: [] }
}

2. Implement Context Analysis

• Parse PR/issue body (extract requirements, acceptance criteria)
• Extract linked issues (PR: Add missing chatmode index entries and create master indexes for instructions and collections #123 references)
• Identify labels (epic, feature, bug, enhancement, etc.)
• Auto-detect plan type from labels/description
• Handle edge cases (closed issues, archived repos, etc.)

3. Implement Plan Generation

Create plan template based on type:

Architecture Plan:

• Design Review (Stakeholder alignment, Design docs)
• API Contract (Request/response schemas)
• Data Model (Database schema, migrations)
• Implementation (Code changes, PRs)
• Documentation (API docs, README)
• Deployment (Release notes, breaking changes)

Implementation Plan:

• Setup (Environment, dependencies)
• Core Logic (Main feature implementation)
• Tests (Unit, integration, e2e)
• Documentation (Code comments, user guide)
• Review (Code review, feedback)
• Merge (PR approval, deployment)

Task Plan:

• Analysis (Understand requirements)
• Research (Investigate solutions)
• Implementation (Code changes)
• Verification (Test, verify)
• Documentation (Update docs)

4. Implement Comment Posting

• Format plan as structured markdown
• Include checklist with [ ] checkboxes (auto-progress-tracking)
• Add context links (linked issues, related PRs)
• Post to PR/issue
• Handle dry-run mode (log instead of post)
• Return comment ID for tracking

Test Plan

# Run tests
npm test -- scripts/agents/__tests__/planner.agent.test.js

# Coverage check
npm test -- --coverage scripts/agents/planner.agent.js

# Manual test (dry-run)
DRY_RUN=true node scripts/agents/planner.agent.js

# Integration test
node scripts/agents/planner.agent.js --apply --pr 123
# Should post plan to PR #123

Effort Estimate

Large (4-6 hours)

Priority

🟡 MEDIUM - Core planner functionality

Dependencies

• Requires: [Add GitHub Copilot MCP configuration with Playwright and GitHub remote servers #1.1], [Add GitHub Copilot MCP configuration with Playwright and GitHub remote servers #1.2], [Add GitHub Copilot MCP configuration with Playwright and GitHub remote servers #1.3], [Add GitHub Copilot MCP configuration with Playwright and GitHub remote servers #1.4] (Phase 1 completed)

Blocks

• None (independent from reviewer improvements)

Related Issues

• Reference: ai/improvement-plan-planner-reviewer.md (Issue 2.1)
• Related spec: agents/task-planner.agent.md
• Related instructions: instructions/planner.instructions.md

References

• scripts/agents/planner.agent.js (current stub)
• agents/task-planner.agent.md (specification)
• ai/audit-planner-reviewer-agents.md (Section 2.1, gap analysis)
• GitHub Actions context: https://docs.github.com/en/actions/learn-github-actions/contexts