Guide · Agent skill
How to use the PR Body Drafter agent skill (pr)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The PR Body Drafter is an agent skill, a folder of instructions that an AI coding agent loads when it recognises a matching task. This guide explains what the pr skill from the oh-my-claudecode project makes the agent do, how to install it, how to trigger it, and where it fits and does not fit. By the end you will know what inputs it expects and what the drafted pull request body will contain.
PR Body DrafterYeachan-Heo/oh-my-claudecode · MIT Download the skill Quick start
- 1Download the ZIP. Download the pr skill ZIP from the skill page on this site. It contains SKILL.md.
- 2Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
- 4Run the verify pass. Complete the oh-my-claudecode verify protocol so there is BUILD, TEST, LINT and FUNCTIONALITY output to consume.
- 5Ask for the PR body. Request the PR description in plain language once the work is verified.
- 6Review the output. Check that the Evidence quotes match your verification output and that the Reversibility call is correct.
What the pr skill makes the agent do
The skill writes a pull request body from the work already recorded in an oh-my-claudecode session. It does not gather new evidence. It assembles a fixed template from the paper trail that earlier steps produced.
The template has three parts: a Summary with a one-line claim and the smallest visual that proves it, an Evidence section with before and after pairs quoted from earlier verification output, and a Reversibility section that says how easily the change can be rolled back.
- Summary: one sentence, plus one visual matched to the change (pseudocode, call tree, component tree, shallow file tree, Mermaid diagram, or a diff).
- Evidence: BUILD, TEST, LINT and FUNCTIONALITY output from the verify protocol, quoted exactly.
- Reversibility: a rollback mechanism if the change is a two-way door, or the reason it is expensive if not, plus a one-line blast radius.
- Language: terms taken from GLOSSARY.md, or from CONTEXT.md in legacy projects not yet migrated.
Install the pr skill
Download the ZIP from the skill page on this site. The skill is a single file, SKILL.md, so the ZIP contains that one file in a folder named for the skill. Install it in one of two ways.
You can unzip it into your user skills folder, ~/.claude/skills/, so it is available in every project. Or you can unzip it into a project's .claude/skills/ folder so it applies only to that repository.
In the Claude apps you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal install: unzip into ~/.claude/skills/.
- Project install: unzip into .claude/skills/ inside the repository.
- Claude apps: upload the ZIP under Settings → Capabilities → Skills.
Trigger the skill with natural requests
You do not need a special command. The skill is meant to run when a session opens or materially updates a pull request, and when the launch workflow's Phase 5 completion report needs a PR body. Ask for the PR body in plain language after the work is verified, for example: "Draft the PR description for this change."
Make sure the verify pass ran first. The skill consumes that output, so a request made before verification has little to work with.
- "Write the PR body for this branch from the verify results."
- "Update the PR description now that the fix passes tests."
- "Draft the reviewer-facing PR text for the completed change, including rollback notes."
Example: what a drafted PR body looks like
The output follows the template in SKILL.md: a Summary heading with one claim and one visual, an Evidence heading with Before and After lines copied from verification output, and a Reversibility heading. The exact wording depends on your session's output, so the skill never invents results.
The output is ready to paste into the pull request description. The skill also lists the paper-trail pointers it relied on, such as verify evidence, ADRs and the spec.
- Output: the PR body, ready to paste.
- Output: the pointers to verify evidence, ADRs and the spec it used.
## 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>What the skill needs and what it does not do
The skill depends on the oh-my-claudecode workflow. It reads the verify protocol output, ADRs, and GLOSSARY.md (or CONTEXT.md). The source file references these artifacts but this page has not verified them, so check that your project actually produces them.
The skill runs no scripts and uses no network, according to the item's assessment. It does not run tests or prove claims. If work was not verified, the PR body can only describe it, not check it.
- Requires: oh-my-claudecode workflow, verify protocol output, ADRs, GLOSSARY.md.
- Runs scripts: no.
- Uses network: no.
- Does not re-collect evidence, so unverified work will not be proven by this skill.
Tips for better PR bodies
Run the verify pass before asking for the PR body, so the Evidence section has real before and after output to quote. For visual changes, the skill prefers a captured screenshot, and it notes that qa-tester can drive the runtime to produce one.
Keep ADRs and GLOSSARY.md current. The Reversibility section links an ADR when the change implements or touches one, and the Language section depends on the glossary.
- Verify first, then draft.
- For UI changes, prefer a screenshot captured during verification.
- Link ADRs for changes that implement or touch them.
- Keep the blast radius line to what else the change can move.
What people use it for
- Writing a reviewer-facing PR body after a multi-step agent-assisted change in the oh-my-claudecode workflow.
- Updating an existing PR description after a material change, using the new verification output.
- Producing rollback notes for a change, classifying it as a two-way or one-way door.
- Keeping PR wording consistent with the project's GLOSSARY.md terms across a team.
Frequently asked questions
Is the pr skill free to use?
This page does not state a licence beyond the item's MIT licence, and it does not list prices. The source repository is oh-my-claudecode by Yeachan-Heo, under MIT. Check the repository for current licence terms.
What do I need before the pr skill works?
You need the oh-my-claudecode workflow with its verify protocol output, plus ADRs and GLOSSARY.md, which the skill references. Without verification output the skill has little evidence to quote.
Does the pr skill run tests?
No. It consumes existing verification output and never re-collects evidence. Work that was not verified will be described but not checked.
Where do I put the skill files?
Unzip the ZIP into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
Does the skill need network access or run scripts?
According to the item's assessment, it runs no scripts and uses no network.
Can I use the pr skill without oh-my-claudecode?
The skill depends on the oh-my-claudecode verify pipeline and ADRs. Projects without them will have little for the skill to consume, so it is not a good fit for them.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the pr skill if you work in the oh-my-claudecode workflow with a verify protocol, ADRs and GLOSSARY.md, and you want reviewers to get consistent evidence and rollback notes. Pick something else if your project lacks those artifacts or if you expect the skill to run tests and prove claims itself.
Download the skill