Agent Skills

A repository of reusable agent skills and the docs site that publishes them.

The source of truth lives in skills/. The published documentation lives in docs/content/docs/. Local harness integrations read the same skills through symlinks in .agents/skills/.

Quick Start

For your project

Skills follow the Agent Skills format and can be installed using the skills CLI.

npm:

npx skills add gohypergiant/agent-skills

pnpm:

pnpm dlx skills add gohypergiant/agent-skills

• Only select skills that are relevant for your project
• We recommend "Project" for the installation scope
• We recommend "Symlink" for the installation method

Browse the published docs locally

The documentation app lives in docs/ and uses pnpm.

cd docs
pnpm install
pnpm dev

Then open http://localhost:3000.

Rebuild local skill symlinks

bash scripts/symlink-agent-skills.sh

Run this after adding a new directory under skills/, or any time .agents/skills/ has gone stale.

What is this repository?

This repo is for creating, maintaining, and publishing agent skills.

Each skill is a directory under skills/ with its own SKILL.md and, in most cases, companion files such as CHANGELOG.md, AGENTS.md, README.md, references, scripts, or assets. The docs/ app turns that material into a browsable Fumadocs site.

If you want the user-facing introduction to skills, start with the docs site content instead of this root README:

• docs/content/docs/index.mdx
• docs/content/docs/getting-started.mdx

Why this repo exists

Skills are easier to maintain when the source files, local harness wiring, and published docs stay close together.

This layout gives you a few practical benefits:

• edit canonical skill content in skills/
• expose those same directories to local agent harnesses through .agents/skills/
• publish matching documentation from docs/content/docs/
• keep repo-specific authoring rules in one place with AGENTS.md

If you're here to add or revise a skill, this repo is the working area. If you're here to learn how to use a skill, the docs app is the better entry point.

Repository Layout

agent-skills/
├── skills/                      # Source of truth for maintained skills
├── .agents/skills/              # Symlinks pointing back to skills/
├── docs/                        # Next.js + Fumadocs documentation app
│   ├── content/docs/            # Published docs content
│   └── package.json             # Docs scripts and dependencies
├── scripts/
│   └── symlink-agent-skills.sh  # Rebuild .agents/skills symlinks
├── AGENTS.md                    # Repo rules for skill authors and agents
├── CLAUDE.md                    # Additional agent-facing context
├── CONTRIBUTING.md
└── README.md

Internal Skills

This repository leverages the following third party agent skills internally:

• humanizer
• ask-questions-if-underspecified
• skill-judge
• bash-defensive-patterns

We recommend installing these globally using skills.sh since they are applicable to multiple projects.

Skill Development Workflow

To scaffold and establish a new skill you can invoke the accelint-skill-manager skill like so:

/accelint-skill-manager <description of skill>. Can you help me refine and complete it?

After creating or significantly modifying a skill, run this 4-step audit loop before considering the work done.

Step 1 — Initial skill-judge audit

Run the skill-judge skill against the completed skill. Apply all suggested improvements before proceeding.

Step 2 — accelint-skill-manager audit

Run /clear to reset context, then run the accelint-skill-manager skill against the skill. Apply all structural and content suggestions before proceeding.

Step 3 — Final skill-judge audit

Run /clear, then run skill-judge again. Apply remaining suggestions. Target grade A or higher (>=108/120).

Step 4 — Frontmatter verification checklist

• name is lowercase, no uppercase letters, no consecutive hyphens, ≤64 chars, matches directory name
• description answers WHAT + WHEN + KEYWORDS, is non-empty, ≤1024 chars
• metadata.version is bumped (major for substantial changes, minor for small fixes)

Optional: Step 5 — Exhaustive skill-creator optimization

⚠️ High-cost, high-rigor step. Use for production-critical skills only.

Run /clear, then run /skill-creator "Optimize [skill-name]. Run full test suite with benchmarks and iterate until grade A.". Follow the skill-creator workflow to review outputs, provide feedback, and approve iterations.

API

This repository is content-first rather than library-first, so the public surface is the directory structure and the scripts you work with.

Skill directories in skills/

These are the maintained skills currently in the repo:

• accelint-ac-to-playwright
• accelint-architecture-doc
• accelint-archive-synthesis
• accelint-design-foundation
• accelint-nextjs-best-practices
• accelint-onboard-agent
• accelint-onboard-openspec
• accelint-persona-review
• accelint-prompt-manager
• accelint-qrspi-apply
• accelint-qrspi-archive
• accelint-qrspi-propose
• accelint-react-best-practices
• accelint-react-testing
• accelint-readme-writer
• accelint-security-best-practices
• accelint-skill-manager
• accelint-tanstack-query-best-practices
• accelint-ts-audit-all
• accelint-ts-best-practices
• accelint-ts-documentation
• accelint-ts-performance
• accelint-ts-testing

Current count: 23 skill directories under skills/.

scripts/symlink-agent-skills.sh

Rebuilds .agents/skills/ so local harness discovery points at the current skill sources.

What the script does:

Step	Behavior
Resolve repo root	Finds the repository root from the script location
Ensure target dir exists	Creates .agents/skills/ if needed
Iterate skills	Walks each directory under skills/
Refresh links	Creates or updates ../../skills/<skill-name> symlinks
Protect conflicts	Skips existing non-symlink paths with a warning

Expected link shape:

.agents/skills/<skill-name> -> ../../skills/<skill-name>

Docs app scripts in docs/package.json

The docs app exposes these main scripts:

Script	Command	What it does
dev	pnpm dev	Starts the Next.js docs app locally
build	pnpm build	Builds the production docs bundle
start	pnpm start	Serves the production build
types:check	pnpm run types:check	Regenerates MDX types, runs Next typegen, and checks TypeScript

Examples

Start the docs site

cd docs
pnpm install
pnpm dev

Use this when you want to preview changes under docs/content/docs/ or verify that new documentation renders correctly.

Type-check the docs app

cd docs
pnpm run types:check

This is the most complete verification command exposed by the docs app. It runs fumadocs-mdx, next typegen, and tsc --noEmit.

Rebuild local harness symlinks

bash scripts/symlink-agent-skills.sh

If you added a new skill under skills/, this makes it visible through .agents/skills/ without copying files around.

Inspect a skill directory

A typical skill directory includes files like these:

skills/<skill-name>/
├── SKILL.md
├── CHANGELOG.md
├── AGENTS.md
├── README.md
├── references/
├── scripts/
└── assets/

Not every skill uses every optional file, but SKILL.md is the anchor.

Further Reading

• docs/content/docs/index.mdx — main docs landing page
• docs/content/docs/getting-started.mdx — installation and usage guidance
• docs/content/docs/*/index.mdx — per-skill published docs
• skills/accelint-skill-manager/assets/skill-template/ — starting point for new skills

License

Apache 2.0. See LICENSE.

Architecture & Development Guides

For deeper technical and contributor context:

• AGENTS.md — repo rules for creating and maintaining skills
• CLAUDE.md — additional agent-facing instructions

Contributing

See CONTRIBUTING.md for contributor setup.

Most changes here fall into one of these buckets:

1. update or add a skill under skills/
2. update the published docs under docs/content/docs/
3. maintain local development tooling such as scripts/symlink-agent-skills.sh