Workflow kit with feature-delivery orchestration, PRD, ADR, implementation plans, wave-based subagents, review and verification gates
Given a scope/topic, scans all ADRs reading ONLY their frontmatter (scope, tags, status, title) and returns a ranked shortlist of prior decisions that likely bear on the current work — without loading any ADR body. Use in plan mode (Default Flow step 4) so the planner links related decisions instead of re-deciding them.
Reads one ADR by id or path and returns a compact digest — status, the decision, and its consequences — without loading the full ADR. Use when a plan or review needs to honor a prior decision but not re-read its whole context.
Reads docs/features.md (the feature index) and returns a compact view of related prior features — their status, owner, and any dependency/convention the current feature should inherit — without loading each feature's full brief. Use in plan/triage to spot dependencies and reuse before drafting.
Reads a feature brief or PRD (docs/features/<FEATURE-ID>.md) and returns a compact digest — problem, scope, requirements, acceptance criteria, open decisions — so the caller can plan or review without loading the full document.
Reads ONE Task block from docs/plans/<FEATURE-ID>-plan.md and returns its full detail (objective, files, steps, verification, stop conditions) ready to embed in a Worker subagent prompt. Use during execute to hand a single task to a Worker without loading the whole plan.
Cria commits seguindo o padrão Conventional Commits em português. Divide mudanças em commits pequenos e atômicos. Use quando quiser fazer commit das alterações.
Create an Architectural Decision Record (ADR) document for AI-optimized decision documentation.
Create a new implementation plan file for new features, refactoring existing code or upgrading packages, design, architecture or infrastructure.
Orchestrate lightweight feature delivery for new or existing projects using feature briefs, optional PRDs, optional ADRs, implementation plans, review gates, ownership boundaries, wave-based subagent launch/handoff, and Integration Coordinator execution for parallel human/AI work.
Skill para revisão automatizada de Pull Requests/Merge Requests (GitHub ou GitLab). Use sempre que o usuário pedir para revisar um PR/MR, analisar mudanças de código, fazer code review, ou avaliar um pull/merge request. Também se aplica quando o usuário mencionar "revisar PR", "code review", "review PR", "analisa esse PR", "olha esse pull request", ou fornecer um link de PR/MR. Inclui análise de padrões da codebase, validação de uso de libs via context7, e geração de comentários prontos para postar (somente o subconjunto aprovado pelo usuário).
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Marketplace multi-plataforma para Cursor, Claude Code, Codex e OpenCode — skills reutilizaveis para entrega, planejamento, review e execucao em qualquer projeto.
Repo: lucasAguiar11/agent-skills
skill-forgeworkflow-kit, figma-to-codefeature-deliverySettings → Plugins → Add marketplace lucasAguiar11/agent-skills → Install workflow-kit
/feature-delivery
/plugin marketplace add lucasAguiar11/agent-skills
/plugin install workflow-kit@skill-forge
/plugin install figma-to-code@skill-forge
/reload-plugins
/workflow-kit:feature-delivery
/workflow-kit:review-plan
codex plugin marketplace add lucasAguiar11/agent-skills
/plugins → workflow-kit → Install workflow-kit
@workflow-kit
@feature-delivery
Dentro deste repositorio as skills ja estao disponiveis automaticamente via opencode.json.
Para usar em outro projeto, adicione os caminhos no opencode.json:
{
"skills": {
"paths": [
"/caminho/para/agent-skills/plugins/workflow-kit/skills",
"/caminho/para/agent-skills/plugins/figma-to-code/skills"
]
}
}
Ou via referencia no opencode.json do projeto:
{
"references": {
"skill-forge": {
"path": "/caminho/para/agent-skills",
"description": "Skills reutilizaveis de entrega e workflow"
}
}
}
Depois de editar o config, reinicie o OpenCode para carregar as skills.
workflow-kitfeature-delivery (orquestrador)quick-plan (investigacao leve: explorar -> gate de design -> plano aprovado, sem artefatos)create-implementation-planupdate-implementation-planreview-plantest-guideverification-before-completionsimplifyprdcreate-architectural-decision-recordcommitpr-reviewsupersede-featureReader agents (Claude Code — auto-discovery em agents/, context offload): plan-reader, plan-detail-reader, feature-reader, adr-reader, adr-correlator, feature-index-reader. Cada um lê um doc grande e devolve um digest de forma fixa, mantendo o thread principal enxuto. Nos demais hosts (Cursor/Codex/OpenCode) os docs são lidos inline.
figma-to-codefigma-to-code — Figma to code quase pixel perfect, independente de stack, com verificacao visualO orquestrador escolhe o menor conjunto de artefatos para a mudança e avança por portões verificáveis. Os passos pesados (waves, subagents) só aparecem quando há trabalho paralelo real; um micro-change usa só Goal / Tasks / Verification / Risks.
triage → classifica o pedido, registra a feature em docs/features.md
│ (precondição de cada modo é auto-guiada: se faltar, para e diz o comando exato)
plan → feature brief / PRD / ADR (quando estrutural) + plano
│ ADR exige `scope` → adr-correlator linka decisões anteriores (não re-decide)
│ plano preenche Traceability (REQ→design→teste) e Libraries (lib+versão+doc)
▼
Validation ┌─ self-check V-001..V-007 ─────────────┐
do plano │ algum fail? → status: needs-resolve │
│ corrige → re-checa │──┐ loop até clean
└───────────────────────────────────────┘◄─┘
│ só vira `planned`/aprovável com status: clean (sem gap, libs com doc ref)
review → findings dentro do plano; pós-execute inclui test-guide (keep/improve/remove/missing)
execute → Integration Coordinator: waves, handoffs, verificação por wave
│ precondição: plano `approved` + Validation `clean`
checkpoint → Post-feature Checkpoint (lixo + checagens por limiar) antes de commit/PR
Portões de validação do plano:
| Check | Pega |
|---|---|
| V-001 | REQ sem Task |
| V-002 | Task sem comando de verificação |
| V-003 | decisão blocking em aberto |
| V-004 | write paths sobrepostos em paralelo |
| V-005 | Worker no Launch Spec sem Task |
| V-006 | requisito sem teste (gap na Traceability) |
| V-007 | lib nova/pinada sem doc ref |
Em docs grandes, o plan/review/execute no Claude Code delega a leitura a um Reader agent (digest), em vez de carregar o arquivo inteiro no contexto.
agent-skills/
├── .agents/plugins/marketplace.json
├── .claude-plugin/marketplace.json
├── .cursor-plugin/marketplace.json
├── opencode.json
└── plugins/
├── workflow-kit/
│ ├── agents/ # Reader agents (Claude Code)
│ └── skills/
└── figma-to-code/
└── skills/
/plugin marketplace update skill-forge
/reload-plugins
codex plugin marketplace update skill-forge
MIT
Detalhes: PUBLISH.md
npx claudepluginhub lucasaguiar11/agent-skills --plugin workflow-kitWorkflow Figma to code quase pixel perfect, independente de stack, com extracao via MCP do Figma, spec com proveniencia, mapeamento de tokens/componentes e loop de verificacao visual
Evidence-gated AI coding workflow: scan → analyze → plan → TDD → execute → fix → verify → review, powered by Codebase Memory MCP >= 0.9.0 with optional Serena LSP intelligence. Includes blast-radius planning, test/cycle gates, independent review, and Windows Git Bash hook auto-resolution.
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Harness-native ECC operator layer - 67 agents, 278 skills, 94 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
v9.54.1 — Reliability wave: tangle contextual review correction loop with hard round ceiling, progress-supervised review rounds (per-agent stall watch, descendant-tree kills), council diversity and agy pin fixes, marketplace generator source-of-truth fix, provider troubleshooting runbook and cost-expectations docs. Run /octo:setup.