원클릭으로
retro
Run a retrospective for the last completed work item. Pulls DORA metrics, summarizes flow, captures keep/stop/start.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Run a retrospective for the last completed work item. Pulls DORA metrics, summarizes flow, captures keep/stop/start.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Behavioral floor management. Propose changes, review proposals, apply approved changes, view status. Routes to COO as behavioral floor guardian.
Compliance program management. Propose changes, review proposals, apply approved changes, audit conformance, view change log.
Generic floor management. Routes to the declared guardian for any governance floor. Use /floor propose <floor-name> <change> to propose changes to any active floor.
Structured agent-to-agent handoff. Validates artifact, logs metrics event, dispatches receiving agent.
Interactive project setup: copy templates, configure fleet, define compliance floor, add first specialist agent.
Product Owner commands -- backlog management, prioritization, grooming, review, and status. Routes to the product-owner agent.
| name | retro |
| description | Run a retrospective for the last completed work item. Pulls DORA metrics, summarizes flow, captures keep/stop/start. |
| argument-hint | [item-id] |
Run a team retrospective as defined in .claude/COLLABORATION.md § Team Retrospective (Phase 8).
/retro -- Retro for the most recently accepted item/retro 42 -- Retro for a specific itemCheck escalations. Run ops/feedback-log.sh check-escalations to process any stale proposals before the retro begins.
Gather metrics. Run ops/dora.sh --item <id> (or ops/dora.sh if no item specified) to pull DORA + flow quality data for this iteration.
Summarize the flow. For the target item, report:
Collect reflections. For each agent that participated in the item, generate a reflection:
Synthesize. SM consolidates reflections into 3-5 actionable proposals. Each proposal should be:
Present to user. Show the metrics summary, agent reflections, and proposals. Ask the user to approve, modify, or defer each proposal.
Apply approved changes. For each approved proposal:
.claude/COLLABORATION.mdfleet-config.jsonCheckpoint. After applying changes, SM runs Phase 9 (Checkpoint): evaluate pace, assess process health, and determine if a pace promotion or demotion is warranted based on updated metrics.
## Retrospective: Item #<id> — <title>
### Metrics
- Lead time: X
- Handoff cycles: X
- First-pass yield: X%
- Bugs: X found, X fixed
- Rework loops: X
- PR review rounds: X
- Time-to-merge: X
- PR comments: X
### Agent Reflections
#### <agent-name>
- Keep: ...
- Change: ...
- Try: ...
### Proposals
1. [proposal] — Metric: [what improves] — Scope: [who/what]
2. ...
### Pace Check
Current: <pace> | Recommended: <pace> | Reason: ...