convez.ai
SkillDesenvolvimento WebIntermediário
kryptobaseddev

Awesome Skills – Registro Curado de Agent Skills

kryptobaseddev/awesome-skills

Registro curado, validado e auto-indexado de skills de agent

Ver no GitHub
InicialPluginScore · de 0 a 100

Coleção validada e auto-categorizada de Agent Skills para Claude Code e outros runtimes, com especificação agentskills.io. Inclui validator config-driven, registry.json em máquina-readable, e pipeline CI/CD automático que mantém o catálogo sincronizado via git hooks e GitHub Actions.

Instalação

$npx convez-ai kryptobaseddev/awesome-skills

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 mês
Versão
—
Issues abertas
1
Contribuidores
2

Casos de uso

  • Descobrir e instalar skills prontos para Claude Code
  • Validar novas skills contra a especificação agentskills.io
  • Manter registry centralizado com CI/CD automático
  • Compartilhar skills modulares entre teams e projetos

Veredito

Repositório bem-estruturado e maduro que resolve o problema real de descoberta e validação de skills em ecossistema de agents. A abordagem config-driven com pipeline automático é prática, embora exija compreensão da especificação agentskills.io para máximo proveito.

README

A tradução para o português deste README ainda está na fila.

Kryptobaseddev's Awesome Skills

A curated, validated collection of Agent Skills for Claude Code and other agent runtimes.

Every skill in this repo is hand-written by @kryptobaseddev, validated against the agentskills.io specification, and categorized for easy discovery. The skills below are auto-generated from each SKILL.md frontmatter — see the pipeline section for how the list stays current.

What's a skill?

A skill is a self-contained folder (name/SKILL.md + optional scripts/, references/, assets/) that teaches an AI agent how to do one thing well. Agents discover skills by reading the description field at startup and load the full SKILL.md only when triggered. See agentskills.io for the open specification.

Install

Pick the install location that matches your agent runtime:

RuntimeSkills directory
Claude Code (global)~/.claude/skills/
Claude Code (project)<project>/.claude/skills/
Codex CLI~/.agents/skills/ or <project>/.agents/skills/

To install one skill:

# Symlink (recommended — pulls future updates from git)
ln -s /path/to/awesome-skills/skills/<skill-name> ~/.claude/skills/<skill-name>

# Or copy (snapshot)
cp -r /path/to/awesome-skills/skills/<skill-name> ~/.claude/skills/

To install all skills at once:

for d in /path/to/awesome-skills/skills/*/; do
  ln -sf "$d" ~/.claude/skills/
done

Repository layout

awesome-skills/
├── README.md                    # this file (skills list auto-generated below)
├── registry.json                # machine-readable index of all skills
├── skill-validator.config.yml   # optional repo-wide validator overrides
├── scripts/
│   └── build_registry.py        # scans skills/, updates README + registry
├── .github/workflows/
│   ├── validate.yml             # runs skill-validator on PRs
│   └── registry.yml             # auto-updates README on push to main
├── .githooks/
│   └── pre-commit               # local registry update on commit
└── skills/
    ├── skill-validator/         # the validator itself (config-driven, general-purpose)
    ├── skill-evaluator/         # runtime quality eval + regression detection
    └── <every-other-skill>/

Pipeline

The registry and this README's skills list update automatically:

TriggerWhat runs
git commit (local).githooks/pre-commit regenerates registry.json and the README skills block. Skip with git commit --no-verify.
Push to main.github/workflows/registry.yml regenerates, commits, and pushes if changes are detected.
PR to any branch.github/workflows/validate.yml runs skill-validator on every changed skill. Fails the PR on errors.

Manual refresh:

python scripts/build_registry.py           # writes README + registry.json
python scripts/build_registry.py --dry-run # show what would change
python scripts/build_registry.py --check   # exit 1 if stale (CI-friendly)

Category resolution

Each skill is assigned to one category. Resolution order:

  1. Explicit override — metadata.category: <slug> in the skill's SKILL.md
  2. Rule-based match — first match in skills/skill-validator/config/categories.yml (keywords, name prefix, tags)
  3. Fallback — "Other" 📦

To add a new category, edit categories.yml and rerun build_registry.py.

Validation

Every skill in this repo passes the skill-validator compliance gauntlet:

  • Tier 1: SKILL.md exists, valid YAML frontmatter, no forbidden fields
  • Tier 2: Field types, lengths, regex patterns; recommended metadata keys (author, version, last_updated)
  • Tier 3: Body under 500 lines, no placeholders, file references resolve
  • Tier 4 (opt-in): Manifest integration
  • Tier 5 (opt-in): Provider compatibility map

The validator is config-driven. Edit skill-validator.config.yml at the repo root to customize rules for this project. See skills/skill-validator/config/README.md for the full schema.

Validate one skill locally:

python skills/skill-validator/scripts/validate.py skills/<skill-name>

Validate every skill:

for d in skills/*/; do
  echo "=== ${d} ==="
  python skills/skill-validator/scripts/validate.py "$d"
done

Contributing a new skill

  1. Fork this repo (or branch if you have write access).
  2. Generate scaffold — copy an existing skill that's close to your concept, rename the directory, and edit the frontmatter.
  3. Required frontmatter — name (matching directory), description (≤1024 chars with at least one trigger phrase like "Use when ..."), metadata.author, metadata.version, metadata.last_updated (format: YYYY-MM-DD HH:MM:SS).
  4. Validate — python skills/skill-validator/scripts/validate.py skills/<your-skill>. Fix every error and warning.
  5. Regenerate registry — python scripts/build_registry.py (or just commit; the pre-commit hook will do it).
  6. PR — the CI runs the validator on every skill change; it'll block on errors.

License

Each skill carries its own license field in its frontmatter. Repo scaffolding (scripts/, GitHub Actions, the README) is MIT.


Skills

17 skills across 12 categories — auto-generated from skills/*/SKILL.md. Last updated: 2026-06-11 23:48:38.

🧰 Skill Development

SkillDescription
plugin-creatorScaffold, validate, test, and ship a production-grade Claude Code plugin + marketplace — especially the pattern where the plugin's ONLY component is an MCP server that surfaces a registry of skills…
skill-evaluatorEval-driven quality evaluation, regression detection, and auto-improvement of Agent Skills. Auto-generates challenging test cases unique to the target skill from its SKILL.md, runs A/B benchmarks (…
skill-validatorValidate any Agent Skill against the agentskills.io specification plus project- or skill-local rule overrides. Use when auditing a skill folder for compliance, preparing a skill for distribution, c…

🌐 Browser & Web Automation

SkillDescription
agent-browserBrowser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing we…

💳 Payments & Commerce

SkillDescription
payment-provider-oauthBuild OAuth connections for payment providers (Stripe Connect, Square) in SvelteKit applications. Use when creating booking platforms, marketplaces, or applications where businesses need to connect…

🎨 Frontend & UI

SkillDescription
svelte5-sveltekitComprehensive guide for building modern web applications with Svelte 5 and SvelteKit. Use when creating Svelte components, implementing Svelte 5 runes ($state, $derived, $effect, $props), building …

🖥️ Infrastructure & Sysadmin

SkillDescription
proxmox-adminUse this skill to administer a Proxmox VE 8.x/9.x host, node, or cluster remotely — provisioning, day-2 ops, troubleshooting, hardening, IaC, migration off VMware/ESXi. Covers KVM VMs (qm), LXC con…

☁️ Cloud & Deployment

SkillDescription
flarectlCloudflare CLI management via the flarectl tool. Use when the user asks to manage Cloudflare zones, DNS records, firewall rules, cache purging, or zone exports using flarectl. Covers onboarding (in…
railwayOperate Railway infrastructure: create projects, provision services and databases, manage object storage buckets, deploy code, configure environments and variables, manage domains and networking, s…
runpodctlRunpod CLI (runpodctl) to manage your Runpod GPU workloads from the terminal. Use when you need to create or terminate GPU pods, list available GPU types and prices, inspect pod status, manage SSH …

🗄️ Databases & ORMs

SkillDescription
drizzle-ormExpert guidance for Drizzle ORM and drizzle-kit, covering the v1.0.0-beta (RQBv2, defineRelations, new migration folder structure, consolidated validators) and stable 0.x releases. Includes Node 24…
neonctlComprehensive Neon CLI (neonctl) management for serverless Postgres. Use when managing Neon projects, branches, databases, roles, connection strings, IP allowlists, operations, or authentication vi…

📧 Communication & Email

SkillDescription
resendUse when working with the Resend email API — sending transactional emails (single or batch), receiving inbound emails via webhooks, managing email templates, tracking delivery events, or setting up…

✨ Code Quality & Refactoring

SkillDescription
code-optimizerDeep code optimization audit using parallel specialist agents. Each agent hunts for performance anti-patterns, inefficiencies, and suboptimal code using pattern-based detection (Grep/Glob) WITHOUT …

🛡️ Security

SkillDescription
security-reviewThreat-model-driven security review of a change, feature, or subsystem. Runs a STRIDE-style pass (Spoofing, Tampering, Repudiation, Info disclosure, Denial of service, Elevation of privilege), exam…

🎬 Media & Content

SkillDescription
image-nanobananaGenerate and edit production-ready, watermark-free image assets with Google's Nano Banana models — gemini-3.1-flash-image (Nano Banana 2) and gemini-3-pro-image (Nano Banana Pro) — via the paid Gem…

💡 Quick Helpers

SkillDescription
btwAsk a quick side question about your current work without derailing the main task. Answers from existing conversation context only — no tool calls, no file reads, single concise response. Use when …

Acknowledgements

Ver o original no GitHub

Bom para

Relacionados