Guide · Agent skill
How to use the Branch PR Helper agent skill (branch-pr)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The branch-pr skill is an agent skill: a SKILL.md instruction file that you add to an AI coding agent so it follows a set of rules when it prepares a branch or pull request. This guide explains what the skill makes the agent do, how to install it, how to trigger it, and what it needs and cannot do. It is written for the Gentle AI repository, so you will be able to open a pull request that matches that project's contribution rules after reading it.
Branch PR HelperGentleman-Programming/gentle-ai · MIT Download the skill Quick start
- 1Download and unzip. Download the skill ZIP from the directory page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
- 2Or upload in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the ZIP.
- 3Name the destination. Tell the agent the target repository and base branch, not your fork.
- 4Make a natural request. Ask the agent to prepare or open a pull request for your change.
- 5Authorize the target read. When asked, give explicit authorization for the exact remote target, operation and credential or session.
- 6Review the draft. Check the policy-derived requirements, draft text and which checklist items remain unchecked.
- 7Perform the human steps. Commit, push, open the PR and handle any merge yourself, since the skill leaves those to you.
What the branch-pr skill makes the agent do
The skill applies when you ask an agent to create, open, or prepare a pull request. Its description reads: creating, opening, or preparing PRs, following the destination repository's contribution policy.
The skill tells the agent to work from the destination repository's own rules. It reads that repository's CONTRIBUTING.md, agent instructions, PR template and documented workflow, and it does not substitute Gentle AI's rules for them. It also resolves the PR base repository rather than a fork's origin.
- Check the destination's policy first; if it requires an issue or approval, verify the linked issue.
- Use only the labels and branch or commit conventions the destination actually defines.
- Draft the PR from the destination's current PR template, or use a concise summary, changes and observed test evidence when no template exists.
- Leave checkboxes unchecked unless there is observed evidence for the claim.
- Stop before committing, pushing, creating a PR, merging, or granting consent; those steps stay with a human.
Install the branch-pr skill
Download the ZIP for the skill from the directory page and unzip it. Place the resulting folder in one of two locations: ~/.claude/skills/ to make it available for your user account, or the .claude/skills/ folder inside a specific project to make it available only in that project.
You can also upload the skill in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the ZIP there.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
How to trigger the skill with natural requests
You do not need a special command. Ask for the task in plain language that matches the skill's trigger, such as creating, opening, or preparing a pull request. The agent then uses the skill when the request fits its description.
Name the repository and the change you want so the agent can resolve the correct base repository and find the related issue, if policy requires one.
- "Prepare a pull request for my fix to the Gentle AI repo. Check the contribution policy first."
- "Open a PR branch for feature X against the upstream base repository, not my fork."
- "Draft the PR description from the project's template and tell me which checklist items I can honestly tick."
What the skill needs before it works
The skill depends on access and on files in the repository. Without them the agent cannot verify policy or current checks.
- Git and GitHub access to the target repository.
- GitHub CLI or equivalent access to read branch rulesets, labels, and CI status.
- Project files such as CONTRIBUTING.md and the PR template in the repository.
- Network use is required: the skill reads remote repository state.
Safety notes: scripts and network access
The skill does not run scripts. Its source says it uses the network, so the agent will contact the remote destination. The skill requires explicit authorization for the exact target host, operation and credential or session before any target read, and it says not to probe ambient credentials.
Labels that are protected, such as status:approved and size:exception, need authenticated permission of MAINTAIN or ADMIN plus a current direct human instruction for that exact action. Required CI is taken from current rulesets and branch protection, and CodeRabbit is optional unless the target policy requires it.
- Expect the agent to ask for authorization before reading the remote target.
- Commit, push, PR creation, merge, and native RDD consent remain human-owned.
- Labels are applied only through the canonical issue-creation workflow, after exact instruction and one bounded attempt.
Tips for getting a clean result
Give the agent the destination repository and base branch explicitly. The skill treats a fork's origin as insufficient proof of the policy that applies.
Read the output as a report rather than an approval. The skill distinguishes recommendations from enforced gates and will not call pending required work merge-ready. Run the project's own checks yourself and report failures honestly.
- Confirm the base repository before drafting anything.
- Review which checklist items are unchecked and why.
- Do not ask the agent to push or merge; the skill deliberately does not do either.
What people use it for
- Opening a compliant pull request to the Gentle AI repository that follows its issue-first and labelling rules.
- Drafting a PR description from a project's template with checklist items marked only where evidence exists.
- Checking which required CI checks are current before a maintainer merges a pull request.
- Preparing a branch name and commit plan that matches a destination's documented conventions.
Frequently asked questions
How do I install the branch-pr agent skill?
Download the ZIP from the directory page and unzip it into ~/.claude/skills/ for your user, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
Is the branch-pr skill free to use?
The skill is licensed under MIT, as listed on this page. Check the license file in the source repository for the full terms.
What does the branch-pr skill require?
It needs Git and GitHub access to the target repository, GitHub CLI or equivalent access to read branch rulesets, labels and CI status, and project files such as CONTRIBUTING.md and the PR template. It does not run scripts but does use the network.
Can the branch-pr skill push or merge my pull request automatically?
No. The skill deliberately prevents the agent from committing, pushing, creating or merging a PR on its own. Those steps are left to a human.
Can I use branch-pr on projects other than Gentle AI?
The skill is designed to follow the destination repository's own policy, but this page is written for Gentle AI. Its issue, label and CI rules are specific to that repository, so review the destination's CONTRIBUTING.md before relying on it elsewhere.
Why did the agent ask for authorization before reading the repository?
The skill requires explicit authorization for the exact remote target, operation and credential or session before any target-host read. It does not probe ambient credentials on its own.
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 branch-pr if you contribute to the Gentle AI repository or maintain it and want an AI assistant to follow the issue-first and labelling rules before drafting a pull request. Pick another approach if you want an agent to push or merge automatically, or if you work on a project whose rules differ from Gentle AI's, since the skill is built around that repository's policies.
Download the skill