convez.ai
SkillAutomaçãoAvançado
JCreatesGH

Claude Skills — Automação e QA para Extensões, Licensing e ServiceNow

JCreatesGH/claude-skills

Automação de loja, licensing offline e QA ServiceNow via Agent Skills do Claude

Ver no GitHub
InicialPluginScore · de 0 a 100

Coleção de skills reutilizáveis para Claude que automatizam tarefas técnicas específicas: geração de assets para Chrome Web Store, sistema de licensing offline com Stripe para software pago, e QA + testes automatizados para ServiceNow. Cada skill é uma pasta auto-contida com instruções que Claude carrega sob demanda.

Instalação

$npx convez-ai JCreatesGH/claude-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
1
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
v1.0.0
Issues abertas
0
Contribuidores
1

Casos de uso

  • Gerar assets de dimensão exata para extensões Chrome (ícones, tiles, screenshots)
  • Monetizar software cliente (extensão, desktop) com licensing offline verificado
  • Revisar e testar mudanças ServiceNow antes de promoção com testes ATF automáticos

Veredito

Skills de produção bem testadas para tarefas especializadas: geração de assets sem dependências externas, licensing robusto offline com Stripe, e automação completa de QA ServiceNow. Excelente para quem já conhece Claude Code e precisa de workflows técnicos prontos para clonar e adaptar.

README

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

claude-skills

A small collection of Agent Skills I built for Claude (Claude Code / Claude Desktop). Each skill is a self-contained folder with a SKILL.md that Claude loads on demand when your request matches what the skill does — plus the scripts, templates, and reference docs it needs to actually do the work.

These are real, battle-tested skills extracted from shipping projects. Where a skill needed a secret, a live URL, or anything account-specific, that value has been replaced with a clearly-marked placeholder — so everything here is safe to read, fork, and adapt. You supply your own keys/links.

SkillWhat it doesBest for
chrome-store-assetsGenerates exact-dimension Chrome Web Store graphics (icon, promo tiles, framed screenshots) on macOS with zero image libraries — pure SVG → PNG via the built-in sips. Includes a full Chrome Web Store listing-field cheat sheet.Shipping a browser extension (or any app) to a store, or making pixel-exact PNGs on a Mac without ImageMagick/node-canvas.
paid-extension-stripe-licensingTurns a client-side app (extension/desktop/downloadable) into a paid product with a free tier and offline-verified signed license keys. The Stripe secret never ships in the client — a tiny Cloudflare Worker fulfills purchases and enforces one-device activation.Monetizing downloadable software where you can't trust the client and don't want to run a heavy backend.
servicenow-updateset-qa-atfReviews a ServiceNow update set against its Jira story (change manifest, code/ACL review, AC traceability, risk score) and then authors deployable ATF tests as ServiceNow SDK (Fluent) code.Pre-promotion QA of ServiceNow changes and auto-generating automated tests for them.
unity-headless-validationCompile-checks Unity C# in seconds without launching Unity (bundled Roslyn), and runs -batchmode tests + scene bakes on a lock-free clone while the editor stays open.Fast edit→verify loops and CI gates on a Unity project, especially with the editor already open.
unity-visual-self-qaRenders every Unity scene and popup to exact-size PNGs headlessly (HUD included), then reviews them by reading + cropping + colour-probing.Letting an agent (or CI) see a Unity game's UI/art without a human screenshotting.

Installing a skill

A skill is just a folder containing a SKILL.md. To use one:

Claude Code / Claude Desktop (personal skill, available everywhere):

# clone, then copy the skill(s) you want into your personal skills directory
git clone https://github.com/JCreatesGH/claude-skills.git
cp -R claude-skills/chrome-store-assets ~/.claude/skills/

Restart your Claude client. The skill activates automatically when a request matches its description — you don't invoke it manually.

Project-scoped (only inside one repo): copy the folder into <your-project>/.claude/skills/ instead.

As a .skill bundle (for upload-based installs): grab a prebuilt bundle from the Releases page, or build one yourself (a .skill is just a zip with the skill folder at its root):

zip -r chrome-store-assets.skill chrome-store-assets -x '*.DS_Store'

The skills

🎨 chrome-store-assets

Generate Chrome Web Store (and general app-store) graphics on macOS without any image library — store icon (128×128), small promo tile (440×280), marquee (1400×560), and framed 1280×800 screenshots — by designing in SVG and rasterizing with the built-in sips tool. Also ships a reference for the Chrome Web Store listing fields (permission justifications, privacy-practices answers, categories, character limits, and the "Title/Summary are locked to the manifest" gotcha).

Use it when you're preparing a Chrome extension (or any app) for store submission, or whenever you need exact-dimension PNGs with text/gradients/embedded images on a Mac and ImageMagick / node-canvas aren't available.

What's inside

  • SKILL.md — the SVG → PNG technique, required asset sizes, and hard-won submission gotchas.
  • scripts/make-promo-tiles.mjs — writes branded SVGs for the icon / small tile / marquee from a small config block, then rasterizes them. Pure Node + sips, no npm install.
  • scripts/frame-screenshots.mjs — turns raw screenshots into store-spec 1280×800 PNGs (branded background, rounded frame + shadow, caption).
  • references/listing-fields.md — ready-to-paste answers for the developer-dashboard Privacy practices tab and every common permission.

💳 paid-extension-stripe-licensing

Monetize a client-side app with a free tier and a paid license using Stripe for payment and offline-verified signed license keys. The design principle: never ship a payment secret in downloadable code. The Stripe secret lives only in a tiny Cloudflare Worker that fulfills purchases (mints + emails a signed key) and enforces one-device activation; the client bundles only a public key and verifies licenses offline with WebCrypto — no network needed to validate, and activation fails open so paying customers are never locked out.

Use it when you're asked to "make this extension/app paid," add a license/upgrade flow, gate features behind a purchase, or set up Stripe fulfillment for downloadable software.

What's inside

  • SKILL.md — the architecture, build order, key format, and a list of correctness gotchas (each one was a real bug).
  • scripts/license-tools.mjs — CLI to generate a signing keypair (keygen), mint keys (issue [--master]), and verify a key.
  • templates/worker.js — the Cloudflare Worker: /success post-checkout page, /webhook backup fulfillment, /activate + /deactivate device binding. Holds the Stripe secret.
  • templates/license-client.ts — the client-side offline verifier + activate/deactivate + free-tier gating.
  • references/deployment-runbook.md — step-by-step wrangler deploy + Stripe wiring.

⚠️ You supply the secrets. The Stripe secret key, webhook secret, and your signing private key are read from Worker env vars / a gitignored secrets/ file — none of them are in this repo. The example payment link and activation URL are placeholders; replace them with your own.

🛠️ servicenow-updateset-qa-atf

End-to-end ServiceNow update set QA review and ATF test generation. Given a Jira story and its ServiceNow update set(s), it produces a pre-promotion review (change manifest, code review, ACL/security pass, completeness/dependency check, acceptance-criteria traceability, risk score with test routing), then authors deployable ATF tests as ServiceNow SDK (Fluent) TypeScript. It complements ATF and Instance Scan; it doesn't replace runtime testing.

Use it when someone supplies a ServiceNow update set (XML export) and/or a Jira story key and wants it reviewed, QA'd, risk-assessed, promotion/CAB-checked, or wants ATF/automated tests generated — even if "ATF" or "review" isn't said explicitly.

What's inside

  • SKILL.md — inputs, the two-phase workflow, and operating principles.
  • scripts/parse_updateset.py — dependency-free parser that builds a change manifest from sys_update_xml records, catches the "container-only export" mistake, and flags risk-relevant artifact types.
  • references/analysis-workflow.md — the full Phase 1 review + Phase 2 ATF instructions, risk rubric, and tunable code-review rule set.
  • references/fluent-atf-api.md — verified ServiceNow SDK (Fluent) ATF API with examples.
  • references/deploy-and-run.md — how to deploy the generated ATF via the ServiceNow SDK and run it.
  • examples/ — a worked review + ATF pair showing the expected shape and depth.

🎮 unity-headless-validation

Validate Unity C# changes from the command line. Two capabilities that together remove the slow, single-instance editor from your inner loop: (1) compile-check in ~10-30s without launching Unity by driving the editor's bundled Roslyn (csc.dll) against the engine reference assemblies — it runs even while the editor is open; and (2) run PlayMode/EditMode tests and editor automation (-executeMethod scene bakes) in -batchmode on an APFS copy-on-write clone with the lock files stripped, so they work alongside your open editor. Generated scenes/assets keep their GUIDs, so they drop back into the real project with references intact.

Use it when you've edited a Unity project's C# and want to know it compiles or passes without clicking into the editor — "does this compile", "run the Unity tests headless", "rebake the scene from the CLI", "set up Unity CI", or any edit→verify loop on a project whose editor is already open.

What's inside

  • SKILL.md — the edit→compile→clone→test→adopt loop and operating principles.
  • scripts/roslyn-compile-check.sh — compile a project (runtime or editor) via Unity's bundled Roslyn; prints the error count and top error classes. Pure bash, no installs. Handles the two non-obvious reference-set gotchas (netstandard shims required; monolithic UnityEditor.dll must be excluded).
  • scripts/clone-unity-project.sh — APFS copy-on-write (or plain) clone with lock/instance/layout files stripped so batchmode runs alongside the open editor.
  • references/headless-runbook.md — every flag, the compile-recipe rationale, and the batchmode failure-mode catalog (frame-wait flakes, WaitForEndOfFrame no-fires, serialized-value-beats-code-default, sceneLoaded cascade, [InitializeOnLoad] hangs).

📸 unity-visual-self-qa

Let an agent see its own Unity game. A PlayMode test renders every scene — camera plus the screen-space HUD, which a naive cam.Render() misses — into an off-screen RenderTexture and writes one exact-resolution PNG per scene in -batchmode. The same public Capture() shoots popups/menus after you open them (by reflection if the test assembly can't reference your UI). Then you actually review the shots: read them, crop into HUD/text/alignment details with nearest-neighbour zoom, and measure real region colours so "black void vs dark backdrop" is a measurement, not a guess.

Use it when you want to verify how a Unity game looks without a human screenshotting — "screenshot every scene", "does the HUD look right", "visual QA/regression for my Unity game", "audit the menus", or checking art/layout after a change. Pairs with unity-headless-validation to capture a freshly-baked scene on the clone.

What's inside

  • SKILL.md — the capture + review loop, the two harness tricks (overlay-canvas reparent; wall-clock settle), and operating principles.
  • scripts/VisualAuditCapture.cs — the PlayMode capture harness; Capture() is public/static so popup tests reuse it.
  • scripts/probe_image.py — Pillow-only screenshot probe: dims, crop (nearest-neighbour zoom), color (mean/min/max RGB of a region).
  • references/review-loop.md — capture command, popup coverage, the look→crop→measure→fix loop, and gotchas (-nographics black frames, capture-order state leaks, editor-only debug-UI contamination).

Both Unity skills use only what Unity already ships (its bundled .NET/Roslyn, -batchmode) plus git/bash and (for the probe) Pillow. No paid APIs, no secrets, nothing account-specific.


A note on safety

These skills generate and run code, talk to paid APIs (Stripe), and touch external systems (ServiceNow, Chrome Web Store). Treat the templates as starting points:

  • Replace every REPLACE_WITH_… / placeholder value with your own.
  • Keep signing keys and API secrets out of version control (each skill says exactly where they belong).
  • The ServiceNow skill does static analysis and is designed to deploy/run ATF against non-production instances only.

License

MIT © Joshua John. Use them, fork them, adapt them. No warranty.

Ver o original no GitHub

Bom para

Relacionados