From crew
Use before declaring implementation work done to separate task-compliance review from code-quality review and make approval criteria explicit.
How this skill is triggered — by the user, by Claude, or both
Slash command
/crew:review-gatesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The user trusts the "done" signal to mean work has been independently checked. Skipping from "code changed" to "done" without review means the user inherits unchecked risk.
The user trusts the "done" signal to mean work has been independently checked. Skipping from "code changed" to "done" without review means the user inherits unchecked risk.
Apply two gates whenever the task is substantial.
Default rule:
Check:
Check:
Use exactly one:
Include:
When the review is substantial, persist it with:
node "${CLAUDE_PLUGIN_ROOT}/scripts/crew.mjs" write-review-result --repo "$PWD" --title "<short title>" ...The user relies on the review result to know what was actually checked. Leaving standards checking implicit means the user cannot tell whether their configured review program was applied. The reviewer should say which standards and skills were actually applied.
If the reviewer finds scope drift or a missing spec boundary, send the task back to the lead rather than silently repairing the assignment — silent repair removes the independent check the user depends on.
npx claudepluginhub alex-radaev/engineering-os --plugin crewGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.