< English | 简体中文 >
One command polishes any GitHub repo's surface — label taxonomy (emoji rating tiers), issue forms, PR template, CI, rulesets, docs. Repo meta & config only, no code logic touched.
Agent-agnostic: works with any skills-enabled agent — Claude Code, Cursor, Copilot, OpenClaw, and more.
Stop pasting the same wall of prompts every time you start a new project. This skill detects the current state of a repository and applies a professional baseline idempotently: run it on a fresh repo, or run it on an old repo to fill in what's missing. It never clobbers your content.
| Module | What it does |
|---|---|
| 🏷️ Labels | Idempotent upsert of a full label taxonomy — P0–P3 priority, impact: security severity, bug, ready to merge... (rating tiers always emoji) |
| 📋 Issue forms | YAML issue forms for bug / feature / question + config.yml |
| 🔀 PR template | Structured pull request template with checklist |
| 🤖 CI workflows | Auto-detects your test framework (Node / Python / Go / Rust) and adds a matching GitHub Actions workflow |
| 🛡️ Rulesets | Branch protection: required PR review, linear history, no deletions, required signatures |
| 📚 Docs | README badges, CONTRIBUTING.md, SECURITY.md, CHANGELOG.md, CODEOWNERS, optional MIT license |
| ✅ Verification | Re-queries GitHub and prints an applied / skipped / failed checklist |
- Surface-only: touches labels, templates, CI configs, rulesets, and docs — never source code logic.
- Auth-first: checks
gh auth statusbefore anything; if you're not logged in, it tells you exactly how to log in (or provide a token). No token → no action. - Repo-aware: detects personal vs. organization repos and
publicvs.privatevisibility, and adapts what it does. - Idempotent: run it twice, get the same result. Existing labels are updated, not duplicated; existing templates are diffed, not overwritten.
- Dry-run first: shows a plan and asks for confirmation before destructive operations (rulesets, branch rules, deletions).
# via the skills CLI (works with any agent)
npx skills add programmingWTF/repo-standardizer
# or via OpenClaw
openclaw skills install git:programmingWTF/repo-standardizer@mainRequires: gh CLI (authenticated), git, and jq + python3 for label encoding.
Just ask your agent:
Standardize the repo programmingWTF/repo-standardizer
Or run modules selectively:
Add issue forms and a PR template to my repo
Fix the labels on this repo to match the standard taxonomy
Protect the main branch with a ruleset requiring review
What happens under the hood?
- Preflight — verify login (
gh auth status), detect repo type (UservsOrganization,PUBLICvsPRIVATE), audit current state. - Plan — show a table of what will be created / updated / skipped.
- Apply — run modules in order, each idempotent.
- Verify — re-query GitHub and report the final checklist.
.
├── SKILL.md # The skill itself
├── templates/ # Reusable templates
│ ├── labels.json # Label taxonomy (color + description)
│ ├── LABELS.md # Rating-label documentation template
│ ├── issue-form-*.yml # Bug / feature / question forms
│ ├── config.yml # Issue template config
│ ├── PR_TEMPLATE.md # Pull request template
│ ├── ci-*.yml # Node / Python / Go / Rust workflows
│ ├── CODEOWNERS # Default owners file
│ ├── CONTRIBUTING.md # Contribution guide
│ ├── SECURITY.md # Security policy
│ ├── CODE_OF_CONDUCT.md # Code of conduct
│ ├── CLAUDE.md # Claude Code guide template
│ ├── AGENTS.md # Generic AI-agent guide template
│ ├── VISION.md # Project vision template
│ ├── CHANGELOG.md # Keep-a-Changelog skeleton
│ ├── README.md # README skeleton (English)
│ └── README.zh.md # README skeleton (简体中文)
├── docs/ # Project docs
│ └── ARCHITECTURE.md # Skill architecture
├── AGENTS.md # AI-agent guidance (root)
├── CLAUDE.md # symlink → AGENTS.md
├── VISION.md # Project vision & direction
├── CHANGELOG.md # Release history
├── THIRD_PARTY_NOTICES.md # Third-party attributions
├── LICENSE
├── README.md # English (default)
└── README.zh.md # 简体中文
Real repositories standardized by this skill — see SHOWCASE.md.
PRs welcome! See CONTRIBUTING.md for conventions. Found a bug? Use the bug report form.
MIT © 2026 programmingWTF & LiGuiyu-AI — the friendly lobster 🦞