Memory Plan
Create, manage, and track implementation plans for multi-session projects.
Instructions
When this skill is triggered, determine the user's intent:
Intent: Show Current Plan
If user wants to see the current plan:
- •Read
.claude/plans/active_plan.md - •Read
.claude/plans/progress.md - •Read
.claude/plans/findings.md - •Display summary:
code
Active Plan: [Plan Name] Status: [In Progress | Blocked | Not Started] Created: [Date] Progress: [X]% ([N] of [M] phases complete) ## Phases Phase 1: Setup (Complete) - Set up project structure - Configure dependencies Phase 2: Core Implementation (In Progress) - [x] Create data models - [x] Implement API endpoints - [ ] Add validation - [ ] Write tests Phase 3: Integration (Pending) - Connect to external services - End-to-end testing ## Recent Findings - API rate limit is 100 req/min (discovered 2024-01-14) - Need to handle timezone edge cases ## Blockers (none) Commands: - "complete [task]" - Mark a task done - "add task" - Add new task to current phase - "add finding" - Record a discovery - "block [reason]" - Mark as blocked - "next phase" - Move to next phase
Intent: Create New Plan
If user wants to create a new plan:
Step 1: Gather Information
code
Let's create an implementation plan. What's the goal of this work? (e.g., "Add user authentication", "Migrate to PostgreSQL")
Step 2: Define Scope
code
Got it: "[Goal]" What are the major phases? I'll help break them down. (Enter phases one per line, or say "help me figure it out")
Step 3: Break Down Tasks For each phase, ask what tasks are needed.
Step 4: Set Success Criteria
code
What does "done" look like for this plan? Success criteria help us know when we're finished: - [ ] All tests passing - [ ] Documentation updated - [ ] Deployed to staging
Step 5: Generate Plan
Create the plan file at .claude/plans/active_plan.md:
markdown
# Active Plan: [Goal] **Created**: [DATE] **Status**: Not Started **Goal**: [Goal description] ## Overview [1-2 paragraph description generated from discussion] ## Implementation Steps ### Phase 1: [Name] - [ ] Task 1.1 - [ ] Task 1.2 ### Phase 2: [Name] - [ ] Task 2.1 - [ ] Task 2.2 ## Success Criteria - [ ] Criterion 1 - [ ] Criterion 2 ## Dependencies - [Required tools, access, or prerequisites] ## Notes - [Additional context]
Intent: Update Progress
If user wants to update task status:
Mark Task Complete
code
User: complete "Create data models" Marked complete: Create data models Phase 2 Progress: 2/4 tasks (50%) Overall Progress: 5/12 tasks (42%) Remaining in Phase 2: - [ ] Add validation - [ ] Write tests
Intent: Add Finding
If user discovered something during implementation:
code
User: /memory-plan finding: The external API requires OAuth2, not API keys Added to findings.md: ### OAuth2 Required for External API > Discovered: 2024-01-15 The external API requires OAuth2, not API keys as originally assumed. Impact assessment: - May need to add OAuth2 library - Consider adding to Phase 3 tasks?
Intent: Manage Blockers
Add Blocker
code
User: /memory-plan blocked: waiting for API credentials Plan marked as BLOCKED Blocker: Waiting for API credentials Added: 2024-01-15 When the blocker is resolved, say "unblock" to continue.
Remove Blocker
code
User: /memory-plan unblock Blocker resolved: Waiting for API credentials Plan status: In Progress Resuming Phase 3: Integration
Intent: Archive/Complete Plan
When work is finished:
code
User: /memory-plan complete Let's verify the success criteria: [x] All tests passing [x] Documentation updated [x] Deployed to staging All criteria met! Archiving plan: - Moving to: .claude/plans/archive/[date]_[plan-name]/ - Creating completion summary Would you like to: 1. Add learnings to knowledge base 2. Create a decision record 3. Start a new plan 4. Done
File Paths
- •Active plan:
.claude/plans/active_plan.md - •Progress tracker:
.claude/plans/progress.md - •Findings:
.claude/plans/findings.md - •Archive:
.claude/plans/archive/
Related Skills
- •
/memory-start- Shows active plan at session start - •
/memory-save- Updates progress when saving session - •
/memory-knowledge- Store learnings from plan execution