Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Commit Push Pr

ASecurity

Commit current work, push the branch, and create a GitHub pull request with a concise title, body, and verification notes.

2 stars
0 votes
0 copies
0 views
Added 9/22/2026
ai-agentspythonrustbashgitfrontenddocumentation

Security Analysis

A100/100

Scanned 9/22/2026

Install to Claude Code

$npx -y skills add jckeen/dotfiles --skill commit-push-pr --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Commit Push Pr?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Commit Push Pr
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jckeen-commit-push-pr/badge)](https://www.skillsdirectory.com/skills/jckeen-commit-push-pr)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: commit-push-pr
description: Commit current work, push the branch, and create a GitHub pull request with a concise title, body, and verification notes.
---

# Commit, Push, PR

Use when the user asks to commit, push, open a PR, ship the current branch, or
make a pull request.

## Workflow

1. Inspect state:
   - `git status --short`
   - `git diff --staged`
   - `git diff`
2. Determine which files belong to the requested change. Leave unrelated user
   changes alone.
3. Finish simplification, intended documentation, changelog, and generated-file
   updates. Run the smallest useful verification affected by the final changes.
4. Stage only relevant files.
5. Commit with a conventional message:
   - `feat: ...`
   - `fix: ...`
   - `docs: ...`
   - `refactor: ...`
   - `test: ...`
   - `chore: ...`
6. Ask which lane the diff requires, then run **that** gate with `--require
   --committed` **after the last commit, before the push**. Supply `--base <ref>`
   when needed to identify the PR base. The gate reviews the committed delta and
   records private evidence bound to that artifact.

   ```bash
   python3 ~/.claude/scripts/review-receipt.py lane --repo . --scope committed
   ```

   - `required_lane: antigravity` (ordinary tier-2 work) →
     `~/.claude/scripts/antigravity-review-gate.sh --require --committed`.
   - `required_lane: codex` (risk surfaces, or an unreadable classification) →
     `~/.claude/scripts/codex-review-gate.sh --require --committed`. Never
     downgradable: an Antigravity receipt cannot ship such a diff, and the gate
     says so when it runs as the supplementary lane (ADR-0008).
   - `required_lane: any` (tier-1 docs diff) → either gate; its tier valve mints
     the exemption receipt without dispatching a reviewer.
   - An Antigravity gate that exits 3 could not run at all — rerun the diff
     through the Codex gate. Exit 2 is a verdict; fix the findings instead.
   - `review-and-push.sh` performs this whole selection itself.
   - Blocking findings, unreadable output, or an unavailable reviewer stop
     shipping. Fix in a follow-up commit and repeat affected verification and
     final review. Do not treat a degraded run as approval.
   - Exit 0 alone does not mean review completed. Distinguish a successful
     review receipt from an explicit tier/no-diff exemption; record an
     exemption as an exemption, never as a reviewer verdict.
   - For the self-instruction guard, independently read the affected
     instructions and route refutation through a different model family.
     A scoped `CODEX_GATE_ALLOW_INSTRUCTION_DIFF=1` override may follow that
     independent review; record the reason and evidence. Alternatively use
     the Antigravity gate with `--require` and its receipt when it provides the
     independent review. Never bypass hooks to evade the guard.
7. Require a review from a different model family than the implementer for
   authentication, authorization, secrets, payments, destructive operations,
   schemas, or public trust boundaries. A fresh Codex reviewer of Codex work
   supplies context independence only. Use a suitable separate-family reviewer
   and verify its actual identity; an Antigravity dispatch label alone does
   not prove the model used. For other relevant runtime/frontend changes,
   `~/.claude/scripts/antigravity-review-gate.sh` can add an advisory opinion;
   run it before step 6, never after. Starting a review in either lane
   retires the other lane's receipt for the same artifact (#480), so an
   advisory run after the required gate voids the receipt about to be pushed.
   A text-diff review is not runtime/browser verification.
8. Immediately before pushing, validate the artifact evidence:

   ```bash
   python3 ~/.claude/scripts/review-receipt.py check --repo . \
     --head "$(git rev-parse HEAD)"
   ```

   No `--reviewer`: the receipt's own recorded classification decides which
   lanes may ship this diff (ADR-0008), so naming a lane there could only
   narrow the check, never satisfy a stronger requirement. Pass the same
   `--base <ref>` if one was selected. A missing, stale,
   mismatched, or below-requirement receipt blocks the push; only a checker-accepted current
   exemption may replace completed review under the applicable gate policy.
   Any subsequent artifact edit invalidates approval. Recommit intended edits,
   rerun affected verification and required reviews, and recheck the receipt.
   Push the current non-default branch, setting upstream if needed.
   For an explicitly selected nondefault PR base, scope that same base to the
   push with `REVIEW_RECEIPT_BASE=<ref> git push ...` so the hook checks the
   intended receipt. Without this one-push setting the hook requires the
   repository's default base; never select a narrower base just to pass it.
9. Create a PR with `gh pr create`:
   - title under 70 characters
   - body covering what changed, why, and how it was tested
   - issue links such as `Fixes #123` when applicable
10. Inspect `gh pr checks` and report pending or failed checks. Require the
    applicable artifact review evidence and required CI checks to be green
    before enabling auto-merge (see ADR-0003).
11. Enable auto-merge only when an applicable standing order explicitly grants
    that authority and the required review/CI conditions are satisfied.
    Otherwise return the PR URL and verification state without merging.

## Safety

- Never stage secrets or generated runtime state.
- Never revert unrelated work.
- Never push implementation directly to a default or protected branch.
- Never force-push, bypass hooks, or amend published history under this skill.
- If verification fails, stop and fix or report the failure before pushing.

## Worktree disposition

Finish delivery by recording the disposition of every task worktree. Follow
`claude/scripts/README.md` → Worktree lifecycle for the shared release,
preview, and retirement commands. Stop its reviewer/runtime processes first
and run lifecycle commands from outside the target worktree.

After verified integration, retire released worktrees only when the user or
applicable standing authorization permits cleanup. Archive recovery data and
review evidence before moving the directory into retained, locked quarantine.
Retirement does not delete the quarantine or reclaim its disk space.
Preserve primary/current/locked
worktrees, unique work, dirty/untracked/ignored content, and stashes. Never use
branch names or commit subjects as merge evidence.

If the PR is pending, record a release for the exact completed artifact when
appropriate, retain the worktree with its owner and PR, and give the next
session the retirement command. A later merge does not grant deletion rights.
The hygiene timer inventories these releases but never removes worktrees.
Unknown ownership or unavailable remote evidence means retained with a reason.

Attribution

jckeenjckeen
View sourceMore from jckeen →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

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 (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1066601 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

651 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →