Skip to main content

​
Overview

Wrap-Up is an end-of-session ritual that ensures you leave your codebase in a clean state and capture learnings for future sessions. It’s different from session-handoff — Wrap-Up is a checklist for you, while handoff is a document for the next session.

​
Trigger

Use when:
  • Ending a coding session
  • Saying “wrap up”, “done for now”, “finish coding”
  • Before closing the editor
  • After completing a feature or task
/wrap-up
/pro-workflow:wrap-up

​
Workflow

The Wrap-Up ritual follows a 5-step checklist:
1

Changes Audit

What files were modified? Anything uncommitted? TODOs left in code?
git status
git diff --stat
Output:
  • List of modified files
  • Uncommitted changes highlighted
  • Any new untracked files
2

Quality Check

Run lint, typecheck, and tests. All passing? Any warnings?
npm run lint 2>&1 | head -20
npm run typecheck 2>&1 | head -20
npm test -- --changed --passWithNoTests
Output:
  • Pass/fail status for each gate
  • Error count and first few errors if any
  • Warning count
3

Learning Capture

What mistakes were made? What patterns worked well?Format as [LEARN] Category: RuleCategories:
  • Navigation, Editing, Testing, Git, Quality, Context, Architecture, Performance
4

Next Session Context

What’s the next logical task? Any blockers? Context to preserve?Output:
  • Next task description
  • Known blockers or dependencies
  • Files to review next session
5

Summary

One paragraph: what was accomplished, current state, what’s next.Output:
  • Concise session summary
  • State of the codebase
  • Confidence level (high/medium/low)
Do not skip any checklist step. Each step serves a purpose for clean session endings.

​
Learning Categories

Wrap-Up captures learnings in these categories:
Code changes, patterns, wrong approachExamples:
  • [LEARN] Editing: Use early returns instead of nested if statements
  • [LEARN] Editing: Always use async/await, not .then() chains
Test approaches, coverage gaps, flaky testsExamples:
  • [LEARN] Testing: Mock external APIs in auth tests to prevent flakiness
  • [LEARN] Testing: Run tests before commit, not after
Commits, branches, merge issuesExamples:
  • [LEARN] Git: Always pull before creating new feature branches
  • [LEARN] Git: Use conventional commit format for consistency
Lint, types, style violationsExamples:
  • [LEARN] Quality: Run typecheck after adding new interfaces
  • [LEARN] Quality: Fix lint warnings immediately, don't accumulate
When to clarify, missing requirementsExamples:
  • [LEARN] Context: Ask about error handling strategy before implementing
  • [LEARN] Context: Clarify scope before touching >5 files
Design decisions, wrong abstractionsExamples:
  • [LEARN] Architecture: Keep business logic out of React components
  • [LEARN] Architecture: Prefer composition over inheritance
Optimization, O(n^2) loops, memoryExamples:
  • [LEARN] Performance: Use indexes for frequently queried fields
  • [LEARN] Performance: Debounce search input, don't query on every keystroke

​
Guardrails

No Skipping Steps

Every checklist item must be completed. If tests are failing, flag before ending session.

Uncommitted Changes

If uncommitted changes exist, ask whether to commit or stash. Never leave in ambiguous state.

Learning Format

Always use [LEARN] Category: Rule format. Keep rules to one line, specific and actionable.

Quality Gate Failures

If lint/typecheck/tests fail, document why and create TODO for next session.

​
Output

Wrap-Up provides a comprehensive session report: 
WRAP-UP REPORT
==============

1. CHANGES AUDIT
   Modified (5 files):
   M  src/api/users.ts
   M  src/types/user.ts
   M  tests/api/users.test.ts
   M  src/middleware/auth.ts
   M  package.json
   
   Uncommitted:
   ⚠️  src/utils/debug.ts (new file, not tracked)
   
   TODOs:
   - src/api/users.ts:42: TODO(#123): add rate limiting

2. QUALITY CHECK
   ✓ Lint: passed (0 errors, 2 warnings)
   ✓ Typecheck: passed
   ✓ Tests: 23 passed, 0 failed
   
   Warnings:
   - src/utils/debug.ts: 'debugMode' is defined but never used

3. LEARNINGS CAPTURED
   [LEARN] Testing: Mock external APIs in auth tests to prevent flakiness
   [LEARN] Navigation: User types are in src/types/user.ts not src/models/user.ts

4. NEXT SESSION CONTEXT
   Next task: Add rate limiting to login endpoint (#123)
   
   Blockers: None
   
   Files to review:
   - src/middleware/auth.ts (modified authentication flow)
   - Check if debug.ts should be committed or removed

5. SESSION SUMMARY
   Completed user profile update endpoint with PATCH /api/users/:id.
   Added validation for email format and username immutability.
   All tests passing. Ready to merge to main after adding rate limiting.
   
   State: HIGH confidence, clean codebase, clear next steps.

==============
Ready to end session?

​
Examples

​
Example 1: Clean Session

User: "wrap up"

Claude:
[Runs full checklist]

WRAP-UP REPORT
==============

1. CHANGES AUDIT
   Modified (3 files):
   M  src/components/Button.tsx
   M  src/components/Button.test.tsx
   M  src/styles/button.css
   
   ✓ All changes committed (last commit: a3f5c21)
   ✓ No untracked files
   ✓ No TODOs

2. QUALITY CHECK
   ✓ Lint: passed
   ✓ Typecheck: passed
   ✓ Tests: 15 passed

3. LEARNINGS CAPTURED
   No new learnings this session.

4. NEXT SESSION CONTEXT
   Next task: Add loading state to Button component
   Blockers: None

5. SESSION SUMMARY
   Refactored Button component to use CSS modules. Improved
   type safety with stricter prop types. All tests passing.
   Ready for code review.
   
   State: HIGH confidence.

==============
Ready to end session?

​
Example 2: Uncommitted Changes

User: "done for now"

Claude:
[Runs checklist]

WRAP-UP REPORT
==============

1. CHANGES AUDIT
   Modified (2 files):
   M  src/api/webhooks.ts
   M  tests/api/webhooks.test.ts
   
   ⚠️ UNCOMMITTED CHANGES
   
   Should I:
   1. Commit these changes with /smart-commit
   2. Stash for later with `git stash save "webhook retry logic"`
   3. Continue wrap-up without committing

[After decision...]

2. QUALITY CHECK
   ✓ Lint: passed
   ✓ Typecheck: passed
   ✗ Tests: 12 passed, 1 failed
   
   Failure:
   tests/api/webhooks.test.ts:45
     Expected status 200 but got 500
   
   ⚠️ Failing test should be fixed or documented as known issue.

3. LEARNINGS CAPTURED
   [LEARN] Testing: Test edge cases before calling wrap-up

4. NEXT SESSION CONTEXT
   Next task: Fix failing webhook test (#45)
   
   Blockers: Test failure needs investigation

5. SESSION SUMMARY
   Added webhook retry logic with exponential backoff.
   Implementation complete but one test failing. Need to
   investigate why webhook returns 500 in test environment.
   
   State: MEDIUM confidence, test failure needs attention.

==============
Ready to end session? (Recommended: fix test first)

​
Example 3: Quality Gate Failure

User: "wrap up"

Claude:
[Runs checklist]

WRAP-UP REPORT
==============

1. CHANGES AUDIT
   Modified (1 file):
   M  src/utils/parser.ts
   
   ✓ All changes committed

2. QUALITY CHECK
   ✗ Lint: 3 errors
   
   src/utils/parser.ts:12:5
     Missing semicolon
   
   src/utils/parser.ts:18:3
     'parseJson' is defined but never used
   
   src/utils/parser.ts:25:10
     Unexpected console.log
   
   ⚠️ Typecheck: not run (lint must pass first)
   ⚠️ Tests: not run (lint must pass first)
   
   Quality gates failed. Should I:
   1. Fix the lint errors now
   2. Create TODO for next session
   3. Amend the last commit after fixing

[After decision...]

3. LEARNINGS CAPTURED
   [LEARN] Quality: Run lint before committing, not after
   [LEARN] Editing: Remove debug console.log statements before commit

4. NEXT SESSION CONTEXT
   Next task: Fix lint errors in parser.ts
   Blockers: Lint errors must be fixed before continuing

5. SESSION SUMMARY
   Added JSON parser utility. Committed but lint errors
   introduced. Need to fix linting before moving forward.
   
   State: LOW confidence, technical debt introduced.

==============
⚠️ Session ending with quality gate failures.
Recommended: Fix lint errors first.

Continue anyway?

​
Integration with Pro Workflow

Smart Commit

Wrap-Up calls Smart Commit if uncommitted changes exist

Session Handoff

Run handoff after wrap-up to generate resume context

Learn Rule

Learnings captured during wrap-up use learn-rule format

Insights

Wrap-Up data feeds into session analytics

​
Configuration

​
Custom Quality Commands

Override default commands in CLAUDE.md: 
## Wrap-Up Quality Gates

Run these before ending session:
- Lint: `npm run lint -- --max-warnings 0`
- Types: `tsc --noEmit`
- Tests: `npm test -- --coverage`
- Security: `npm audit`

​
Skip Specific Checks

Temporarily skip checks (use sparingly): 
# Skip tests (if blocked by external service)
WRAP_UP_SKIP_TESTS=true /wrap-up

# Skip lint (not recommended)
WRAP_UP_SKIP_LINT=true /wrap-up

​
Best Practices

Make it a habit. Even 5-minute sessions benefit from wrap-up discipline.
Fresh context makes better learnings. Don’t wait until later.
All changes committed, tests passing, no surprises for next session.
If ending with issues, clearly document what’s blocking and why.

​
Troubleshooting

​
Wrap-Up Takes Too Long

# Run only fast quality gates
npm run lint -- --quiet
npm test -- --onlyChanged

​
No Learnings to Capture

That’s fine! Not every session has mistakes. But consider:
  • What worked well that should be repeated?
  • What was faster than expected?
  • What was slower and why?

​
Quality Gates Always Failing

If quality gates consistently fail:
  1. Fix technical debt before new features
  2. Add pre-commit hooks to catch earlier
  3. Adjust thresholds if too strict

​
Next Steps

Try Session Handoff

Generate resume context for smooth session continuity

Master Learn Rule

Turn corrections into permanent memory

View Insights

Analyze patterns from wrap-up data

Explore All Skills

See the complete skill system

Build docs developers (and LLMs) love

Get started for freeTalk to us