/retro

name: retro
description: Sprint retrospective — analyzes git velocity, commit patterns, test health, and shipping cadence. Use after a sprint, at the end of the week, or when the user wants to reflect on progress.
argument-hint: "<days-back>"
allowed-tools: Bash, Read, Grep, Glob

Sprint Retrospective

Retrospectives turn shipping data into actionable insights. This skill analyzes your git history, commit patterns, test health, and shipping velocity to surface what's working and what needs attention.

**Announce at start:** "I'm running a sprint retrospective on this project."

Process

Step 1: Gather Data

Run the retro analyzer:

node ${CLAUDE_PLUGIN_ROOT}/tools/retro-analyzer.mjs <project-directory> --days 7

For custom date ranges:

node ${CLAUDE_PLUGIN_ROOT}/tools/retro-analyzer.mjs <project-directory> --since 2026-03-24

Step 2: Present the Retrospective

Format the data as a clear, actionable report:

Velocity Report Card

+===========================================+
|     S P R I N T   R E T R O              |
+===========================================+
|  Period        Mar 24 – Mar 31           |
|  Commits       47                         |
|  Lines Added   2,341                      |
|  Lines Removed 892                        |
|  Net Change    +1,449                     |
|  Authors       1                          |
+===========================================+

Commit Breakdown

Show the distribution of commit types (features, fixes, refactors, tests, docs, chores) as a visual breakdown. Highlight the feature-to-fix ratio — a healthy project ships more features than fixes.

Hot Files

List the most-changed files. Files with excessive churn may need to be split into smaller modules.

Test Health

• Number of test files
• Passing / failing tests
• Whether a test script exists
• Recommendation if tests are absent or failing

Shipping Cadence

• Peak day of week (when do you ship most?)
• Peak hour (when are you most productive?)
• Tags/releases in this period

Step 3: Insights & Recommendations

Based on the data, surface:

**What went well:**

• High velocity periods
• Good deletion-to-addition ratio (code hygiene)
• Consistent shipping cadence
• Test coverage improvements

**What needs attention:**

• Fix-heavy sprints (more fixes than features = quality debt)
• Hot files that churn excessively
• Missing or failing tests
• Uneven shipping cadence (feast-or-famine pattern)

**Action items:**

• Concrete, specific recommendations with clear next steps
• Link to relevant Ultraship skills (e.g., "Use `/tdd` for the untested module")

Step 4: Cross-Reference with Learnings (Optional)

If the project has learnings (from `/learn`), search for relevant context:

node ${CLAUDE_PLUGIN_ROOT}/tools/learnings-manager.mjs search --query "sprint"
node ${CLAUDE_PLUGIN_ROOT}/tools/learnings-manager.mjs search --query "debugging"

Incorporate past learnings into recommendations.

Multi-Project Mode

For founders running multiple projects, run retros across all of them:

node ${CLAUDE_PLUGIN_ROOT}/tools/retro-analyzer.mjs ~/project-a --days 7
node ${CLAUDE_PLUGIN_ROOT}/tools/retro-analyzer.mjs ~/project-b --days 7

Compare velocity across projects to understand where time is being spent.

Output

The retrospective should be concise and actionable — not a data dump. Lead with insights, support with data. The goal is to help the user ship better next week.