Simiancraft Skills
simiancraft/simiancraft-skills
Skills prontos para produção: planejamento, testes e padrões de composição
Ver no GitHubColeção curada de skills e agents para Claude Code, desenvolvidos internamente pela simiancraft e prontos para produção. Inclui metodologias de planejamento tático, padrões de composição React, harnesses de teste para Android e web, e automação de integração.
Instalação
npx convez-ai simiancraft/simiancraft-skills1 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
Casos de uso
- Planejar refatorações complexas com metodologia hands-off
- Testar aplicações Android com emulador headless e Maestro
- Automatizar testes de integração web com Playwright
- Estruturar componentes React com padrão de zona e data ownership
- Testar segmentação de imagem com MediaPipe em emulador
Veredito

Coleção sólida e opinada de skills de produção, com foco em planejamento tático e harnesses de teste. Qualidade garantida (roda em produção na simiancraft), mas escopo específico para times de engenharia que trabalham com React, Android e automação de integração.
README
A tradução para o português deste README ainda está na fila.
simiancraft-skills
Curated Claude Code skills and agents. Only what we run in production.
Opinionated Claude Code skills and agents from simiancraft. Quality-bar-only.
What you get
A planning methodology that produces something like this, cold-handoff-ready and disposable by design:
# Split monolithic resource-actions hook into per-action files
**Status:** Draft
**Scope:** subsystem
**Date:** 2026-04-15
**Last reviewed:** 2026-04-15
**Context:** A single 870-line hook serves five mutations; one mutation's logic
has broken another's twice in the last quarter.
## Goal
Replace the monolith with a per-action structure where each mutation owns its
own home, composed by a thin dispatcher that preserves today's consumer-facing
shape. Done = no monolith remains; consumer call sites are byte-identical to
today; existing test bodies pass without modification.
## File structure: after
**Legend:** 🆕 new · ✏️ rewritten
```
features/resources/
├── ✏️ use-resource-actions.ts // dispatcher only, ~60 lines
└── 🆕 actions/
├── 🆕 create.ts // extracted from use-resource-actions.ts
├── 🆕 rename.ts // extracted from use-resource-actions.ts
└── 🆕 ...
```
## Commits
### Commit 1: Scaffold actions/ folder
**Files created:** `features/resources/actions/{create,rename,...}.ts`
**Gate:** Project full validation passes.
### …
### Commit N+1: Delete this plan
- Delete `features/resources/split-resource-actions-hook.md`.
**Gate:** Project validation passes. Repo contains no references to the plan file.
The plan file gets deleted as the last commit of the work it described. Two-key handshake before deletion. 150-word Goal cap. Atomic commits with verification gates.
Install
/plugin marketplace add simiancraft/simiancraft-skills
/plugin install simiancraft-skills@simiancraft-skills
Update later:
/plugin marketplace update simiancraft-skills
What's in here
Skills
how-to-plan: methodology for tactical, hand-off-ready planning docs. Goal-as-north-star, atomic commit steps with verification gates, before/after file trees with a 10-symbol legend, and the Inspector Gadget Rule.zone-composer: React composition pattern for domain features (screens, panels, tools, editors). A chassis owns data and state branching via flat early returns; presentational leaves take named zones instead of flag props; mutations live inactions/; layouts are polymorphic. Lean core plus six task references (key patterns, polymorphic layouts, FSM wizards, refactoring, GraphQL fragments, code style).android-emulator-harness: bring up an Android app in a headless emulator on Linux/WSL and drive it for automated integration testing, the Android analog of Playwright for web. Boots an AVD under KVM, installs an APK, drives the UI with Maestro, and gates on logcat plus screenshots.android-emulator-mask-testing: specialization of the harness for camera and segmentation testing. Puts a real person in front of the emulator camera so MediaPipe / ML Kit selfie segmentation produces a mask, then verifies background-replacement, blur, and shader effects. Ships a pre-framed subject fixture.playwright-harness: the operational trunk for browser work, the web analog of the emulator harness. Writes a script, launches Chromium (real GPU under WSLg via ANGLE when WebGL matters), drives the page, and gates on screenshots plus collected page errors. Two references carry the depth:interactions.md(locators, actions, waits, assertions) andflows.md(login and session reuse, forms, responsive sweeps, link-checking, network stubbing). Specializations layer on top.playwright-camera-mask-testing: specialization of the harness for camera and segmentation testing in the browser. Feeds a real person through getUserMedia so MediaPipe selfie segmentation produces a mask, then verifies background-replacement, blur, and shader effects by vision. Ships a webcam-shaped subject fixture and an empty-scene negative case.playwright-gif-capture: specialization of the harness for capturing an animated GIF of a page, canvas, or WebGL animation. Drives the clock deterministically, grabs frames, and encodes a looping GIF, with a reference on the color, frame-rate, and size tradeoffs so the output is not deep-fried, janky, or oversized.
Agents
android-emulator-tester: automated Android UI/integration testing specialist; drives a real app on a headless emulator and gates on what it observes. Owns the boot, install, drive, and assert loop, and is honest about what an emulator cannot validate (real FPS, true camera/mic fidelity, network quality).android-kotlin-expert: Android native specialist for Kotlin, Java, Gradle, the Jetpack libraries, OpenGL ES and camera pipelines, and React Native / Expo Modules native bridging. Two modes: implement or review.
Curation policy
Deliberately small, slow-growing. Every skill is one we use in production and are willing to put the simiancraft name on; nothing makes it in for completeness or volume. See CONTRIBUTING.md for the contribution posture.
License
MIT. See LICENSE.