new-branch-and-pr

An agent skill by cursor, from cursor/plugins. Tags: delivery, git, github, version-control, workflow.

What it does

Create a fresh branch, complete work, and open a pull request

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add cursor/plugins --skill new-branch-and-pr

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/cursor/plugins
cp -r plugins/cursor-team-kit/skills/new-branch-and-pr ~/.claude/skills/new-branch-and-pr

Safety box score

Grade B, low risk: 21 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This is a documentation-only git workflow skill: it tells the agent to create a branch, make changes, commit them, push, and open a pull request. The main risks are remote repository mutation and sending local work to a git host; there are no bundled scripts, downloads, or secret-handling steps.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	1 (low)	The skill requires the agent to "Complete implementation and tests" and then "Commit focused changes and push," which implies shell/git command execution, but it ships no scripts or code of its own.
Network access	2 (moderate)	It explicitly says to "push" changes and "Create a concise PR," which sends repository contents and metadata to a remote git service.
Filesystem reach	1 (low)	The workflow operates on the repo working tree and git metadata only: "Ensure the working tree is clean" and "Create a descriptive branch from the latest main."
Secrets & credentials	0 (none)	There is no request to read, paste, or use tokens, keys, passwords, dotfiles, or other credential stores.
Privilege & persistence	0 (none)	The skill does not mention sudo, global installs, services, scheduled tasks, or OS/shell configuration changes.
Instruction hijack surface	0 (none)	No instruction asks the agent to ignore safeguards, hide actions, follow remote instructions, or adopt a deceptive persona.
Irreversible actions	2 (moderate)	It directs the agent to independently carry out a multi-step change flow—"Complete implementation and tests," then "Commit focused changes and push" and "Create a concise PR"—which mutates a repo and opens a PR without a step-by-step human gate.
Transparency	0 (none)	The skill contains only plain markdown text; there are no binaries, blobs, or fetched/decoded runtime code.

Before installing

Confirm you are comfortable with the agent creating branches, commits, and PRs in this repository.
Check your remote branch protections and review rules before allowing pushes.
Try the skill first in a throwaway clone to see what branch name, commit(s), and PR text it generates.

Rated 2026-10-19 by an automated review of 1 file from the skill's repository.

Source

Repository: cursor/plugins (all skills from this repository)
Path: cursor-team-kit/skills/new-branch-and-pr/SKILL.md
Branch: main
Collection: cursor
Updated: 2026-09-23

fix-merge-conflicts — Resolve merge conflicts non-interactively, validate build and tests, and finalize conflict resolution
principle-never-block-on-the-human — Apply when tempted to ask 'should I do X?' on reversible work. Proceed, present the result, let the human course-correct after the fact.
principle-outcome-oriented-execution — Apply during planned rewrites and migrations with explicit phase boundaries.
principle-sequence-verifiable-units — Apply to multi-step work (sweeps, migrations, runs of similar edits) and to how you stack commits and PRs.
review-and-ship — Review the current branch for bugs, intent fit, and test coverage; run or write tests; commit focused work; open or update a PR.
caveman-commit — Write a Conventional Commits message compressed to intent only. Use for "write a commit", "commit message", /commit or /caveman-commit.