Cherry Studio Skills — Workflows de Automação Empresarial
CherryHQ/skills
Skills prontas para automatizar testes, PRDs, conteúdo e operações sem perder controle
Ver no GitHubConjunto de skills reutilizáveis do Cherry Studio que automatizam fluxos operacionais repetitivos: testes com rastreamento dual, redação de PRDs, transformação de transcrições em conteúdo e gestão de reembolsos. Projetadas para eliminar "taxa de coordenação" entre ferramentas, mantendo decisões críticas sob controle humano.
Instalação
npx convez-ai CherryHQ/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
- Automação de testes com filing simultâneo em múltiplos sistemas (GitHub + trackers internos)
- Geração de PRDs estruturados bilíngues com pesquisa e gate de aprovação
- Conversão de transcrições em posts ou artigos com correção de ASR e fact-checking
- Processamento de reembolsos com workflow de aprovação e rastreamento financeiro
Veredito

Excelente referência de padrões de automação orientados ao contexto empresarial real (dual-track, approval gates, grounding em corpus). Requer adaptação para sua stack específica, mas os workflows são bem documentados e imediatamente replicáveis. Particularmente valioso para equipes que precisam sincronizar dados entre múltiplas ferramentas sem perder rastreabilidade.
README
A tradução para o português deste README ainda está na fila.
About This Repository
These are the workflows the Cherry Studio team runs every day. We package them as skills so our AI agents handle the boring stuff — browser testing, filing bugs across two systems, drafting PRDs, turning transcripts into content, processing expenses.
We're sharing the ones we've cleaned up enough to be useful outside our walls. We'll keep adding more as we make them shareable.
What These Solve
Most of these cut "coordination tax" — the handoffs between tools and people that burn time but don't need human judgment.
| Pattern | Example | What the skill automates |
|---|---|---|
| Test → Track in two systems | Find a staging bug → file in Feishu Bitable + GitHub simultaneously | Captures browser evidence (DOM, console, network, screenshots), writes both records, deduplicates against existing issues |
| Requirement → Structured PRD → Issue | A feature idea from a meeting note → bilingual draft → GitHub issue | Enforces research (search existing issues, PRs, code), locks format (Background/Goal/Spec/Verification/Related), requires human approval before creation |
| Transcript → Publishable content | A recorded talk → blog post or thread | Learns style from past writing, fact-checks against sources, corrects ASR errors |
| Invoice → Reimbursement → Ledger | A stack of receipts → approval workflow → financial record | Batch aggregation, approval routing, dashboard |
The pattern: humans decide, the skill formats, tracks, and syncs across systems.
Skills
🧪 Testing & Quality
| Skill | Purpose | What it does |
|---|---|---|
| test-and-report | Manual testing + dual-track bug filing | Captures browser evidence, files simultaneously to Feishu Bitable and GitHub, enforces dedup, prompts for priority |
📝 Product & Content
| Skill | Purpose | What it does |
|---|---|---|
| prd-creator | Bilingual PRD → GitHub issue | Research phase, opinionated format, human approval gate, Project board integration |
| transcript-to-content | Transcripts → posts or articles | Style learning, source fact-checking, ASR correction, multi-format output |
💼 Operations
| Skill | Purpose | What it does |
|---|---|---|
| expense-reimbursement | Invoice → reimbursement → ledger | Batch aggregation, approval workflow, financial dashboard |
🎯 Career & Job Hunting
| Skill | Purpose | What it does |
|---|---|---|
| cv-expert | JD-matched resume optimization | 7-dimension diagnosis, Gap Q&A to mine real experience, Word-comment annotations, 校招 / 社招 modes |
| interview-retro | Post-interview retrospective | Per-question analysis, 5-dimension scoring, better-answer templates, next-round prep checklist → Feishu doc |
For External Use
These skills reference our internal stack (Feishu, Cloudflare R2, specific repos) because that's where we run them. The workflow patterns are what's reusable:
- Dual-track submission — filing simultaneously in a project tracker and a dev tracker, with dedup
- Approval-gated issue creation — requiring human review before an agent creates a GitHub issue
- Style-grounded content generation — grounding output in a corpus of past writing
Adapt them to your stack. More skills coming.