agentskills.codes
AG

agentflow-retro

Run a retrospective after every 10 completed User Stories. Use when the retro counter reaches 10, when the team wants to reflect on recent work, or when continuous improvement analysis is needed.

Install

mkdir -p .claude/skills/agentflow-retro && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14755" && unzip -o skill.zip -d .claude/skills/agentflow-retro && rm skill.zip

Installs to .claude/skills/agentflow-retro

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Run a retrospective after every 10 completed User Stories. Use when the retro counter reaches 10, when the team wants to reflect on recent work, or when continuous improvement analysis is needed.
195 chars✓ has a “when” trigger

About this skill

Retrospective — Continuous Improvement

Triggered every 10 completed User Stories (tracked in project memory).

When to Run

  • Retro counter reaches 10 completed User Stories
  • Either party requests: "Let's do a retro"
  • After a significant incident or quality issue

Counter tracks US completions only — spikes, bugs, and tooling tasks don't count.

Retro Template

## Retro -- [date] -- after story N

### What worked well
-

### Friction / slowdowns
-

### Bugfix quality (were root causes documented?)
-

### Memory + tooling gaps
-

### Inner loop robustness
-

### Quality gate effectiveness
-

### Actionable items
- [ ] (add to BACKLOG#Ideation or update project rules)

Process

  1. Gather data — review DONE-Today and archives since last retro
  2. Walk each section — be specific, reference actual tasks/commits
  3. Identify patterns — what keeps recurring?
  4. Generate actionable items — each must be concrete and assignable:
    • Process improvements -> update project CLAUDE.md rules
    • Tool gaps -> add to BACKLOG#Ideation
    • Anti-patterns -> add to KNOWN_PATTERNS.md
  5. Reset counter — set retro_stories_since_last: 0 in project memory

Scope

The retro covers:

  • Workflow friction and cycle time
  • Bugfixing approach quality (were root causes documented?)
  • Memory file usefulness
  • Inner/outer loop robustness
  • Tooling gaps
  • Quality gate effectiveness

Output

  • Actionable items added to BACKLOG#Ideation
  • Rule updates applied to project CLAUDE.md
  • New anti-patterns added to KNOWN_PATTERNS.md
  • Counter reset in project memory

Search skills

Search the agent skills registry