Skip to main content

/retrospective

Run a session retrospective to analyze skill invocations, detect anti-patterns, and generate improvement recommendations.

Usage​

/retrospective              # Analyze current session
/retrospective --full       # Full analysis with all details
/retrospective --export     # Export analysis to file
/retrospective --project PILOT  # Project-scoped analysis (ADR-159)

What This Command Does​

  1. Extracts all skill invocations from the session
  2. Classifies outcomes (success, partial, failed)
  3. Detects anti-patterns (retries, confusion, incomplete)
  4. Calculates effectiveness scores per skill
  5. Generates prioritized improvement recommendations
  6. Updates skill-learnings.json for future sessions

When to Use​

  • At end of significant work sessions (30+ min)
  • After skill failures or confusion
  • During optimization sprints
  • Before major deployments

Output Example​

════════════════════════════════════════════════════════════════════
📊 SESSION RETROSPECTIVE ANALYSIS
════════════════════════════════════════════════════════════════════
Session: 2025-01-01-1430

Metrics:
  - total_skill_invocations: 8
  - successful_invocations: 6
  - failed_invocations: 1
  - overall_success_rate: 0.75

Skill Scores:
  ✅ git-workflow: 100%
  ✅ code-editor: 90%
  ⚠️ deployment: 50%

Anti-patterns Detected: 2
  - excessive_retries: Multiple retries indicate unclear instructions
  - context_confusion: Context confusion indicates poor skill scoping

Improvement Recommendations: 3
  [high] deployment skill has 50% success rate
  [medium] Add clearer examples to affected skills
════════════════════════════════════════════════════════════════════

Implementation​

# The command runs the session-retrospective hook
python3 hooks/session-retrospective.py --manual --latest
CommandPurpose
/optimize-skillsView skill health dashboard
/cxCapture session context
/cxqQuery past learnings

Success Output​

When retrospective completes successfully:

✅ COMMAND COMPLETE: /retrospective
Session: <session-id>
Skills analyzed: N
Success rate: X%
Recommendations: M
Learnings saved: skill-learnings.json

Completion Checklist​

Before marking complete:

  • Session data loaded
  • Skill outcomes classified
  • Anti-patterns detected
  • Recommendations generated
  • skill-learnings.json updated

Failure Indicators​

This command has FAILED if:

  • ❌ No session data available
  • ❌ skill-learnings.json write failed
  • ❌ Analysis script error
  • ❌ Zero skills tracked in session

When NOT to Use​

Do NOT use when:

  • Session just started (no data yet)
  • Only trivial interactions occurred
  • Running in CI/CD (no session context)

Anti-Patterns (Avoid)​

Anti-PatternProblemSolution
Skip at session endLost learningsAlways run before exit
Ignore recommendationsRepeating mistakesReview and apply fixes
Run too frequentlyIncomplete dataRun at session end

Principles​

This command embodies:

  • #9 Based on Facts - Uses actual invocation data
  • #1 Recycle, Extend, Re-Use - Improves existing skills
  • #10 Research When in Doubt - Analyzes patterns

Full Standard: CODITECT-STANDARD-AUTOMATION.md

Status: Active Invocable: Yes