Draft a pull request description from existing project records and verification results. Present the change with a compact visual, before-and-after evidence, rollback details, and its wider impact so reviewers can assess the work.
Draft a PR body from OMC's paper trail — the smallest visual that proves the change, evidence consumed from the verify protocol (never re-collected), reversibility from the ADR test on the diff, glossary language throughout.
Review the instructions the agent will receive before you install.
When work lands behind a review, the PR body is the one artifact a reviewer outside the session reads. Everything it needs already exists in the paper trail — this skill assembles it and adds nothing that was not already established. The summary-visual discipline is inspired by show-me (credited in the frontmatter).
## Summary
<one-line claim, then the smallest visual that proves it>
## Evidence
- **Before:** <output / failing check>
**After:** <output / passing check>
## Reversibility
<Rollback: cheap via <mechanism> | expensive because <reason>>
<Blast radius: one line>
State the change in one sentence, then show the smallest view that makes the point. Match the visual to the change:
Show the whole block only when most of it is new, omitted context would hide ownership, or the reviewer needs a copyable target shape. One visual usually suffices; never all forms.
Consume the verify protocol's output — BUILD, TEST, LINT, FUNCTIONALITY evidence, freshness rule inherited. Present before/after pairs: the exact check that failed before and passes after, quoted from the output itself. When the change is visual, prefer a captured screenshot; qa-tester can drive the runtime for one. Never re-collect evidence here and never claim evidence the verify pass did not produce.
Run the ADR test on the diff itself — the same three questions launch applies to decisions: hard to reverse? surprising without context? the result of a real trade-off? A no on all three is a cheap rollback: say by what mechanism (revert the merge, toggle the flag). Any yes means an expensive rollback: say why. Classify the door first: a two-way door (undo is cheap and ordinary) needs no ceremony beyond naming the undo; a one-way door says so explicitly and names what makes the step irreversible. Link the ADR when the change implements or touches one. Close with the blast radius in one line — what else this change can move.
Terms come from GLOSSARY.md where a glossary exists (legacy yards not yet migrated keep it in CONTEXT.md). The PR is read outside the session; the glossary is the only vocabulary the reader shares with it.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Write or revise instructions, specifications, tickets, and skill files so an agent can act without prior chat history. Make rules checkable, explain their reasons, define completion criteria, and keep authoritative information in one place.
Simplify bloated AI-generated code by removing dead code, duplicated logic, and unnecessary layers while checking that existing behavior stays intact. Use a limited cleanup scope, verify changes with relevant checks, or request a review without edits.
Review module relationships and change history to identify structural improvements in a codebase. Receive ranked candidates with file references, suggested changes, and risks, plus a recommendation for where to start, without modifying code.
Send questions, review requests, or implementation tasks to a supported local AI command-line tool through one shared command. The wrapper handles provider-specific options and saves the result as a Markdown file for later reference.
Turn an unclear idea into linked decision tickets on a repository’s issue tracker. Work through one decision per session, record outcomes and dependencies, and prepare a brief defining the objective, scope, and exclusions for delivery.
Turn a short product idea into working code through requirements analysis, technical planning, implementation, testing, and review. It coordinates agents across these stages, fixes detected problems, and reports assumptions that still need a human decision.