Sprint Retrospective

Analyze commits, work patterns, and code quality with persistent trend tracking.

Best for: End of sprint, weekly review
Time estimate: 10-15 minutes
Skills used: meow:retro
Agents involved: analyst (cost data), documenter (saves report)

Overview

The meow:retro skill analyzes your git history to produce a structured retrospective. It's team-aware (breaks down per-person contributions) and tracks trends across runs to show improvement or regression.

Step-by-step guide

Step 1: Run the retrospective

/meow:retro

Step 2: Review the analysis

The skill analyzes:

• Commits by type (feat, fix, refactor, docs) and author
• Work patterns — velocity, review turnaround, deploy frequency
• Code quality — test coverage trends, security findings over time
• Per-person — contributions, strengths, growth areas (with praise)
• Cost — token usage from memory/cost-log.json via the analyst agent

Step 3: Track trends

Each retrospective is saved, so the next one compares:

• Are we shipping more or fewer features?
• Is review turnaround improving?
• Are security findings decreasing?
• Is cost per feature going up or down?

How the analyst agent contributes

The analyst provides cost and pattern data from memory/:

• Token usage trends (are we using cheaper models when appropriate?)
• Pattern frequency (recurring issues that should become rules)
• Lessons learned that inform the retrospective's recommendations

Next workflow

→ Architecture Decisions — document architectural choices