---

description: Review session history, decisions, and project progress over time. Use for weekly reviews, standups, retrospectives, or understanding past work.

Memory Review

Review and analyze session history, decisions, and project progress over time.

Instructions

When this skill is triggered, determine the user's intent:

Intent: Recent Activity Overview

Default view when no specific request:

Memory Review

Period: Last 7 days (2024-01-08 to 2024-01-15)

## Sessions: 8 recorded

| Date | Duration | Focus | Key Outcome |
|------|----------|-------|-------------|
| Jan 15 | 2.5h | Auth refactor | Completed token refresh |
| Jan 14 | 1.5h | Auth refactor | Fixed race condition |
| Jan 13 | 3h | Auth refactor | Started Phase 2 |
| Jan 12 | 2h | Bug fixes | Resolved 3 issues |

## Decisions: 3 made

- #0005: Exponential backoff for retries (Jan 14)
- #0004: Token bucket rate limiting (Jan 12)
- #0003: PostgreSQL for session storage (Jan 10)

## Knowledge: 2 updates

- authentication.md: Added rate limiting section
- api-patterns.md: Added retry patterns

Commands:
- "details [date]" - Show specific session
- "decision [number]" - Review a decision
- "this week" / "last week" / "this month"
- "search [query]" - Find in history
- "standup" - Generate standup summary
- "retro" - Generate retrospective notes

Intent: Time-Based Review

This Week / Last Week / This Month

Generate summary for the specified period including:

• Session count and total time
• Main focus areas
• Accomplishments
• Decisions made
• Discoveries
• Blockers encountered

Intent: Generate Standup Summary

If user asks for standup:

## Standup Summary

**Yesterday**:
- Completed token refresh implementation
- Fixed race condition in auth middleware
- Added retry logic with exponential backoff

**Today**:
- Continue Phase 2: Token rotation
- Write tests for refresh flow
- Start Redis connection pooling

**Blockers**:
- None currently

Intent: Generate Retrospective

If user asks for retro/retrospective:

## Retrospective: Auth Refactor

### What Went Well
- Clear plan helped track progress
- Decision records captured rationale
- Finding about race condition saved debugging time later

### What Could Be Improved
- Underestimated complexity of token rotation
- Should have checked API rate limits earlier

### Action Items
- [ ] Add API limit checking to research phase
- [ ] Create template for concurrent code review

### Key Learnings
- Always check for race conditions in token handling
- Token bucket is better than fixed window for rate limiting

Intent: Review Specific Decision

If user asks about a specific decision:

1. Read the decision file
2. Show full content with formatting
3. Note if it has been superseded

Intent: Search History

If user wants to find something:

Delegate to /memory-search skill.

File Paths

• Sessions: .claude/memory/sessions/
• Decisions: .claude/memory/decisions/
• Knowledge: .claude/memory/knowledge/

Related Skills

• /memory-search - Search across memory files
• /memory-status - Current state overview