Guide · Agent skill

How to use the PR Body Writer agent skill (pr)

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

The PR Body Writer is an agent skill called pr, from the mattpocock/skills collection (MIT licensed). It tells your AI coding agent to write pull request descriptions in one fixed template: a Summary with a visual, an Evidence section, and a Merge Danger section. This guide explains what agent skills are, how to install this one, how to trigger it, and where it falls short, so you can use it on your next pull request.

PR Body Writermattpocock/skills · MIT Download the skill

Quick start

1Download the ZIP. Get the pr skill ZIP from this directory page and unzip it. It contains agents/openai.yaml, CREDITS.md and SKILL.md.
2Choose where to install. Place the pr folder in ~/.claude/skills/ for personal use, or in a project's .claude/skills/ folder to share it with a repository.
3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill there instead.
4Gather your evidence. Run your tests and capture before and after output or screenshots, since the agent will not run them for you.
5Ask for a PR body. Make a request that mentions writing a pull request body or description, and include the change and whether it can be undone.
6Review the output. Check the Summary visual, the Evidence section and the door and blast radius against what you know before posting.

What an agent skill is and what the pr skill makes the agent do

An agent skill is a folder containing a SKILL.md file with instructions that an AI agent reads when a request matches the skill's description. The pr skill's description is "Use when writing a PR body." When you ask the agent to write a pull request description, it follows the template in SKILL.md instead of its default style.

The template has three parts. The Summary uses the smallest visual that makes the key point clear, such as pseudocode, a call tree, a component tree, a file tree, a Mermaid sequence diagram, or a diff sketch. The Evidence section lists a before and after, such as a screenshot, command output, or a failing and then passing test run. The Merge Danger section states whether the change is a two-way door (easy to undo) or a one-way door (hard to reverse), gives a one-word blast radius, and notes possible ramifications of merging.

It is a writing template. It does not change code, run tests or capture screenshots.
It asks the agent to keep prose brief and skip preambles.
It refers to a GLOSSARY.md file for the user's domain language.

Install the pr skill from a ZIP file

Download the ZIP for the pr skill from this directory page and unzip it. The package contains agents/openai.yaml, CREDITS.md and SKILL.md. You can install it for yourself or for one project.

For a personal install, place the unzipped pr folder inside ~/.claude/skills/. For a project install, place it inside the project's .claude/skills/ folder so teammates who clone the repository get it too.

In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

Personal: ~/.claude/skills/pr
Project: .claude/skills/pr inside your repository

Trigger the skill with a natural request

The agent loads the skill when your request matches its description, so you do not need a special command. Ask in plain language that mentions a pull request body or description. The source does not list other trigger phrases, so if the agent does not pick up the skill, name it directly in your request.

Give the agent the context it needs: the change, what you tested, and whether the change can be undone. The template asks for all of these, and the agent will fill in gaps with less certainty.

"Write the PR body for these changes."
"Write a pull request description for the login refactor, with a before and after and a merge danger section."
"Draft the PR body. This only changes the cache logic and the change is easy to revert."

Example of the Summary visuals

For logic changes, the skill favours pseudocode. For a change to how components are arranged, it favours a tree. When most of a change is new, or the reviewer needs a target shape to copy, it can show a whole block of code. A diff is used when the point is what changes inside an existing shape.

The skill says to place each visual next to the short text it supports and to avoid overloading the reader. You may get one visual or several, depending on what the change needs.

Logic: pseudocode
Control flow: call tree
UI structure: component tree
File responsibility: shallow file tree
Component or data flow: Mermaid sequence diagram
What changes in an existing shape: diff

on(save)
  if content is unchanged
    return cached result
  write new content
  return fresh result

Requirements, scripts and network use

The skill lists no requirements. It does not run scripts and does not use the network. Its only files are the instructions, a credits file, and an agents/openai.yaml file.

The skill is a writing template. Evidence is only as good as what you provide. The template asks the agent to show a before and after, but the agent does not run tests or take screenshots itself. Run your tests and capture output yourself, then give the results to the agent.

Requirements: none listed
Runs scripts: no
Uses network: no
Licence: MIT
Credits: the Summary visuals and placement guidance come from Dex Horthy's show-me skill, copied into pr rather than linked as a dependency

Tips for getting useful PR descriptions

Tell the agent the true state of your tests. If you have failing output before the change and passing output after, paste both, because the Evidence section depends on them.

Be honest about reversibility. The Merge Danger section works best when you say whether the change touches data, public interfaces or other consumers. A one-word blast radius is easier to review than a vague one.

Paste real command output or test results into the chat.
Name the consumers or screens a change might affect.
Ask for one visual, not all of them, if the change is small.
Check that the generated door and blast radius match your understanding before you post.

What people use it for

Writing a consistent PR description for a team that reviews many small changes
Stating rollback risk and blast radius for a change to shared code or a public interface
Producing a Mermaid sequence diagram or call tree to explain a control-flow change to reviewers
Summarising a file-layout refactor with a shallow file tree

Frequently asked questions

How do I install the pr agent skill?

Download the ZIP from this page and unzip it. Put the pr folder in ~/.claude/skills/ for personal use or in your project's .claude/skills/ folder for a shared install. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.

Does the pr skill need any requirements or software?

The skill lists no requirements. It does not run scripts and does not use the network, so it only supplies instructions to your agent.

Is the pr skill free and what is its licence?

The skill is licensed under MIT. The source does not list pricing, so check the directory page for any availability details.

Does the skill run my tests or take screenshots?

No. It tells the agent what to write in the Evidence section, but it does not run tests or capture screenshots. You need to provide the before and after results yourself.

What is the difference between a one-way and two-way door in the pr template?

A two-way door is a change you can walk back easily, so it is lower risk. A one-way door is a destructive or hard-to-reverse change. The Merge Danger section asks you to state which one applies.

When should I not use the pr skill?

Avoid it if your project does not use pull requests or already has its own PR template, since the fixed format may clash with local conventions. It is also a poor fit for non-technical users who only need a plain summary, because the template assumes code, diagrams and test output.

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 open pull requests and want consistent descriptions that state rollback risk, blast radius and a visual summary. Skip it if your project has its own PR template or if you need the agent to run tests or capture evidence, since it only writes the description.

Download the skill