Plan Management

Plan Management Skill

This skill provides expertise in creating and managing structured feature implementation plans. Plans track multi-phase feature development with subagent assignments, dependencies, and progress tracking.

Plan Location and Structure

• Active Plans: Docs/Plans/plan-*.json
• Completed Plans: Docs/Plans/Completed/
• Template: Docs/Plans/plan-template.json
• Update Utility: Docs/Plans/plan-update.py
• Creation Command: .claude/commands/plan.md (invoked via /plan)

Quick Reference: Plan Update Commands

For Agents Working on Plans

Use these commands to update plans efficiently without parsing the full JSON:

# Check current status - what's in progress, what's next?
python Docs/Plans/plan-update.py <plan-file> --status-check

# See phases assigned to you
python Docs/Plans/plan-update.py <plan-file> --my-assignment <agent-name>

# Start the next available phase (checks dependencies)
python Docs/Plans/plan-update.py <plan-file> --start-next

# Complete the current in_progress phase
python Docs/Plans/plan-update.py <plan-file> --complete-current

# Complete with notes explaining what was done
python Docs/Plans/plan-update.py <plan-file> --complete-current --completion-notes "Implemented webhook handler with retry logic"

# Complete with notes and effort tracking
python Docs/Plans/plan-update.py <plan-file> --complete-current --completion-notes "All tests passing" --actual-effort 4.5

# View full summary
python Docs/Plans/plan-update.py <plan-file> --summary

Detailed Updates

# Start a specific phase
python Docs/Plans/plan-update.py <plan-file> --phase 2 --status in_progress

# Update completion percentage
python Docs/Plans/plan-update.py <plan-file> --phase 2 --completion 75

# Record actual effort hours
python Docs/Plans/plan-update.py <plan-file> --phase 2 --actual-effort 6.5

# Complete a specific step within a phase
python Docs/Plans/plan-update.py <plan-file> --phase 2 --step 3 --status completed

# Multiple updates in one transaction
python Docs/Plans/plan-update.py <plan-file> --phase 2 --status in_progress --completion 50 --actual-effort 3

Agent Workflow for Executing Plans

1. Before Starting Work

# Check what's assigned to you
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --my-assignment supabase-backend-specialist

# Or check overall status
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --status-check

2. When Starting a Phase

# Start the next available phase (recommended - checks dependencies automatically)
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --start-next

# Or start a specific phase
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --phase 2 --status in_progress

3. During Work - Update Progress

# Update completion as you progress
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --phase 2 --completion 50

# Mark individual steps complete (auto-updates phase completion %)
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --phase 2 --step 1 --status completed
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --phase 2 --step 2 --status completed

4. When Completing a Phase

# Complete the current phase (recommended - finds in_progress phase automatically)
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --complete-current

# Complete with notes explaining what was accomplished
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --complete-current --completion-notes "Implemented RLS policies and tested with service role"

# Complete with notes AND effort tracking (recommended)
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --complete-current --completion-notes "All migrations applied successfully" --actual-effort 3.5

# Or complete a specific phase with notes
python Docs/Plans/plan-update.py Docs/Plans/plan-feature-name-MM-DD-YY.json --phase 2 --status completed --completion-notes "Webhook handler deployed" --actual-effort 5.5

5. Auto-Archive on Plan Completion

When ALL phases are marked completed:

• The plan is automatically moved to Docs/Plans/Completed/
• metadata.completed_at is set
• metadata.status is set to "completed"
• metadata.total_actual_effort is calculated

Phase Index for Quick Lookups

Every plan includes a phase_index field that's auto-updated:

{
  "phase_index": {
    "total": 4,
    "current_in_progress": 2,
    "completed": [1],
    "blocked": [],
    "not_started": [3, 4],
    "next_available": 3
  }
}

This allows quick status checks without parsing the full phases array.

Creating New Plans

Use the /plan slash command:

/plan Implement real-time notifications for transaction updates

The /plan command will:

1. Research existing code and patterns (Step 0)
2. Analyze the feature request
3. Generate 3-5 implementation phases
4. Assign appropriate subagents
5. Calculate effort estimates with multipliers
6. Create dependency graph
7. Define rollback strategy
8. Save to Docs/Plans/plan-<feature>-MM-DD-YY.json

Plan Structure Overview

{
  "metadata": {
    "created": "ISO-8601",
    "plan_id": "UUID",
    "version": "2.1",
    "status": "active|completed"
  },
  "phase_index": { /* quick lookup */ },
  "planning": { "goal", "reason", "scope", "priority" },
  "exploration_summary": { /* research findings */ },
  "phases": [
    {
      "number": 1,
      "name": "Phase Name",
      "status": "not_started|in_progress|completed|blocked",
      "completion_percentage": 0,
      "assigned_subagent": "agent-name",
      "steps": [ /* individual tasks */ ],
      "dependencies": [ /* phase numbers */ ],
      "deliverables": [ /* outputs */ ]
    }
  ],
  "subagent_assignments": [ /* who does what */ ],
  "rollback_strategy": { /* how to undo */ },
  "dependency_graph": { "ascii": "Phase 1 -> Phase 2" }
}

Status Transitions

Valid status transitions (enforced by plan-update.py):

• not_started -> in_progress or blocked
• in_progress -> completed or blocked
• blocked -> in_progress
• completed -> (none, final state)

Use --force to override transitions if needed.

Dependency Validation

The --start-next command automatically:

1. Finds the first not_started phase
2. Checks if all dependencies are completed
3. Only starts if dependencies are satisfied
4. Reports blocking phases if not

Best Practices

1. Always use plan-update.py - Never manually edit plan JSON
2. Mark in_progress before starting - Visibility for other agents
3. Update completion incrementally - Track progress as you work
4. Record actual effort - Improves future estimates
5. Complete phases promptly - Enables dependent phases to start
6. Use --status-check first - Understand current state before acting

Troubleshooting

"Phase X blocked by dependencies"

The phase depends on other phases that aren't completed yet. Check --status-check to see blocking phases.

"No phase currently in_progress"

No phase has status: in_progress. Use --start-next to begin the next phase.

"Phase X is already in_progress"

Complete the current phase before starting a new one, or use --force to start anyway.

Plan not auto-archiving

Ensure ALL phases have status: completed. Use --summary to verify.

See Also

• plan.md - Plan creation command
• plan-template.json - Full template reference
• plan-update.py - Update utility source
• agent-index.md - Available subagents