/omni:retro-apply
settle a retro PR's kept lessons
What it does
Settles every kept lesson of retro PR n on the PR's own branch, so each ends in an outcome a person approves by merging: it appends one rule to the playbook slot it belongs in and records became, or records staysHere with the rule that already says it, or asks one question on the retro PR and records nothing. It rewrites the PR's body with a line per lesson, the rule quoted word for word with its slot, or the reason, then Closes #<issue>. With retro.guards on it also writes the guard test, proved red then green; off, it writes no code. It refuses, in one line, a retro PR already merged or closed, and never merges. The Omni app queues it for a runner when a retro keeps a lesson, and the retro PR body carries the same line.
When to use it
Use it when a retro PR holds a kept lesson with no outcome and no runner took the job.
How to use it
/omni:retro-apply --retro <n>Example
/omni:retro-apply --retro 2300→ the retro PR's body quoting each new rule with its slot, or its reason, and omni check retro green
Who runs it
You type it in your coding agent.
Its full instructions, written for the agent: kit/plugin/skills/retro-apply/SKILL.md