Skip to content
Back to skills

Pr

ASecurity

Create, update, format, babysit, resolve conflicts and review findings on, and merge Apex pull requests and PR stacks. Use when the user asks to open or update a PR or merge request, align a stack, get CI green, address review or Bugbot findings, resolve merge conflicts, or merge.

  • 314 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 20, 2026
ai-agentsgobashgitapi

Works with

  • cursor
  • terminal
  • api

Security analysis

A100/100

Scanned September 20, 2026

npx -y skills add pensarai/apex --skill pr --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pr?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Pr
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/pensarai-apex/badge)](https://www.skillsdirectory.com/skills/pensarai-apex)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: pr
description: Create, update, format, babysit, resolve conflicts and review findings on, and merge Apex pull requests and PR stacks. Use when the user asks to open or update a PR or merge request, align a stack, get CI green, address review or Bugbot findings, resolve merge conflicts, or merge.
argument-hint: "[create | update | stack | babysit | comments | conflicts | merge | PR number]"
allowed-tools: Bash(gh *) Bash(git *) Bash(bun *)
---

# PR

Apex pull-request workflow. Read this file, then **read every applicable playbook** under [`.agent-instructions/skills/pr/`](../../../.agent-instructions/skills/pr/) before acting. Do not improvise past a playbook.

This is repository-development guidance for coding agents working on Apex. Keep it in the repo's `.agents/skills` tree only; do not copy or link it into `.claude/skills`, `.skills`, `skills`, `~/.agents/skills`, or `~/.pensar/skills`, because Apex scans those locations as runtime skills.

## Invariants

- Resolve `OWNER/REPO`, the target PR's current head/base SHAs, and stack topology from live Git and GitHub state. A body saying "stack" is not proof of dependent branches.
- Ordinary Apex work branches from and targets `canary`. Do not target `main`, retarget a PR, or rewrite stack topology unless the user explicitly asks for that exact change.
- GitHub reads and writes go through `gh api` (REST or GraphQL). Do not use `gh pr create`, `gh pr comment`, or `gh pr merge`.
- Preserve bot-managed body regions (`CURSOR_SUMMARY`, `codesmith:footer`) byte-for-byte when editing human-authored sections.
- Never merge, squash, close, enable auto-merge, force-push, or delete a branch unless the user explicitly authorizes that exact action. A request to resolve conflicts and push does not authorize merging.
- Work from a clean, isolated checkout when the user's current checkout has unrelated changes. Never stash, reset, or overwrite user work to make a PR task convenient.
- Judge CI, reviews, and mergeability on the latest PR head SHA. Pending work or green checks on an older SHA are not done.

## Route

| Intent                                               | Playbook                                                                                  |
| ---------------------------------------------------- | ----------------------------------------------------------------------------------------- |
| Create or update a feature/fix/docs PR               | [`create-and-update.md`](../../../.agent-instructions/skills/pr/create-and-update.md)     |
| Create, inspect, format, or refresh a PR stack       | [`stacks.md`](../../../.agent-instructions/skills/pr/stacks.md)                           |
| Stay until CI and mergeability settle                | [`babysit.md`](../../../.agent-instructions/skills/pr/babysit.md)                         |
| Review comments, Cursor Bugbot, or thread resolution | [`comments-and-bugbot.md`](../../../.agent-instructions/skills/pr/comments-and-bugbot.md) |
| Resolve conflicts or perform an authorized merge     | [`conflicts-and-merge.md`](../../../.agent-instructions/skills/pr/conflicts-and-merge.md) |

After creating, updating, refreshing, or fixing a PR, babysit the new SHA. If it belongs to a stack, also apply `stacks.md`. If review findings exist, apply `comments-and-bugbot.md` before declaring the PR ready.

**Babysit** means staying until required checks have conclusions on the latest SHA, mergeability is known, and no unresolved actionable review work remains. Waiting for human approval after every technical gate is green is terminal; pending CI is not.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.