# Software Engineering Ultimate > Conjunto consolidado de capacidades para transformar o Claude em um agente de engenharia de software, arquitetura, segurança, testes, browser, APIs, banco, IA, DevOps, SEO, performance, documentação e orquestração. Use esta habilidade quando a tarefa exigir execução completa, análise profunda, desenvolvimento, revisão e validação ponta a ponta. - Skill: `sameque00-source/software-engineering-ultimate` (Agent Skill) - Install (CLI): `npx skillmds@latest add sameque00-source/software-engineering-ultimate` - Raw SKILL.md: https://api.skillmd.com/api/skills/sameque00-source/software-engineering-ultimate/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Marketing & Growth - Author: sameque00-source (https://skillmd.com/u/sameque00-source) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/sameque00-source/software-engineering-ultimate --- # Software Engineering Ultimate Esta é uma habilidade consolidada. Execute as instruções abaixo como um sistema único e coordenado. Nunca invente ferramentas, agentes, resultados, integrações ou testes. Quando uma capacidade não estiver disponível no ambiente, registre a limitação e use a melhor alternativa real. Priorize segurança, qualidade, verificabilidade e conclusão ponta a ponta. --- ## CATALOGO DE AGENTES DO PACOTE - **3d**: Graficos 3D: WebGL, Three.js, shaders, cena, camera, iluminacao, performance de render. Use apenas quando a tarefa envolve 3D de verdade. - **accessibility-specialist**: Use this agent when the task involves Audit and improve WCAG-oriented semantics, keyboard access, focus, contrast, labels, and motion. Examples: - **ai-architect**: Use this agent when the task involves Design AI provider abstraction, model routing, prompts, tool use, memory, evals, and fallbacks. Examples: - **api-architect**: Use this agent when the task involves Design and review HTTP APIs, contracts, versioning, validation, error models, and compatibility. Examples: - **architecture**: Analisa arquitetura, dependencias e impacto ANTES de mudar codigo. Use para descobrir quem depende de que, o que quebra junto e se ja existe solucao no projeto. Somente leitura. - **backend**: Implementa e corrige backend: API, rotas, servidor, banco, autenticacao, jobs. Use quando a mudanca e no lado servidor. - **ci-cd-engineer**: Use this agent when the task involves Build reliable CI/CD pipelines, gates, artifacts, rollback plans, and environment promotion. Examples: - **cli**: Ferramentas de linha de comando e scripts de automacao (bash, PowerShell, Python). Use para criar ou corrigir script operacional. - **cloudflare-engineer**: Use this agent when the task involves Design and validate Cloudflare Pages, Workers, routes, bindings, and deployment configuration. Examples: - **coding**: Escreve e refatora codigo de proposito geral. Use quando a tarefa e implementar ou reestruturar codigo que nao se encaixa em backend, frontend ou 3D. - **coordinator**: Consolida os resultados JA produzidos por varios agentes numa sintese unica com veredicto. Use NO FIM, depois que os especialistas entregaram - nunca no inicio (para planejar use queen-coordinator). - **data-quality-engineer**: Use this agent when the task involves Validate data contracts, normalization, invariants, duplication, and corruption risks. Examples: - **database-engineer**: Use this agent when the task involves Design and review schemas, migrations, indexes, queries, transactions, and data integrity. Examples: - **debugger**: Investiga e corrige erro concreto: stack trace, comportamento errado, teste falhando. Use quando ja existe um sintoma observado. - **dependency-auditor**: Use this agent when the task involves Audit dependencies, lockfiles, licenses, versions, supply-chain risks, and unused packages. Examples: - **devops**: CI/CD, deploy, containers, pipeline e configuracao de ambiente. Use para automacao de build e entrega. NAO toca em producao sem aprovacao explicita. - **docs**: Escreve e atualiza documentacao: README, docs tecnicos, comentario de API. Use depois que o codigo esta estavel. - **e2e-engineer**: Use this agent when the task involves Own browser-level end-to-end coverage, fixtures, journeys, screenshots, and regression checks. Examples: - **frontend**: Implementa e corrige frontend: UI, componentes, estado, responsividade, acessibilidade. Use para mudancas no lado cliente. - **hooks**: Hooks do Claude Code e automacao por gatilho de sessao (SessionStart, PostToolUse, etc). Use para configurar ou diagnosticar hook. - **incident-responder**: Use this agent when the task involves Diagnose failures systematically, contain impact, identify root cause, and verify recovery. Examples: - **infrastructure-engineer**: Use this agent when the task involves Review local/runtime infrastructure, processes, ports, environment, and reproducible setup. Examples: - **integration**: Conecta modulos e sistemas: contratos entre camadas, formato de dados, compatibilidade de interface. Use quando frontend e backend precisam conversar. - **mcp**: Servidores MCP, ferramentas MCP e protocolo. Use para configurar, diagnosticar ou estender integracao MCP. - **memory**: Memoria e contexto entre sessoes: o que guardar, onde guardar, como recuperar. Use para organizar conhecimento persistente do projeto. - **observability-engineer**: Use this agent when the task involves Design logging, metrics, tracing, health checks, alerts, and operational diagnostics. Examples: - **optimizer**: Aplica otimizacoes JA identificadas e medidas. Use depois que performance apontou o gargalo - nao use para procurar gargalo. - **performance**: Caca gargalo real de CPU, memoria, rede e render. Use quando algo esta lento. Mede antes de opinar. Somente leitura. - **privacy-engineer**: Use this agent when the task involves Review collection, retention, exposure, consent, minimization, and privacy-sensitive flows. Examples: - **product-architect**: Use this agent when the task involves Define product requirements, acceptance criteria, scope, and user journeys before implementation. Examples: - **prompt-engineer**: Use this agent when the task involves Improve system prompts, task decomposition, agent instructions, evaluation prompts, and guardrails. Examples: - **queen-coordinator**: Planeja a divisao de uma tarefa grande em fases e agentes. Use NO INICIO, antes de despachar qualquer especialista, para descobrir o que pode rodar em paralelo e em que ordem. Entrega um plano de execucao - nao executa nem despacha. - **release-manager**: Use this agent when the task involves Coordinate release readiness, changelog, versioning, gates, rollback, and final sign-off. Examples: - **requirements-analyst**: Use this agent when the task involves Turn vague requests into explicit requirements, constraints, edge cases, and verification criteria. Examples: - **research**: Levanta opcoes, compara alternativas e busca precedente antes de decidir. Use no inicio, quando ainda nao esta claro qual caminho seguir. Somente leitura. - **reviewer**: Revisor adversarial independente. Use por ULTIMO, antes de integrar o trabalho dos outros agentes. Procura o que passou despercebido e questiona decisoes. Nao produz codigo. TEM PODER DE VETO. - **security-auditor**: Varredura ampla e somente-leitura em busca de vulnerabilidades no projeto inteiro. Use para auditoria periodica, nao para revisar uma mudanca especifica (para isso use security). - **security**: Caca exploit, falta de validacao server-side, exposicao de credencial e escalada de privilegio. Use SEMPRE que a mudanca tocar autenticacao, permissao, dinheiro, dado pessoal ou operacao administrativa. TEM PODER DE VETO. - **seo**: Conteudo textual, meta tags, estrutura semantica e copy. Use para escrever o texto da pagina e otimizar descoberta. - **swarm**: Topologia do swarm, distribuicao de carga e coordenacao entre agentes do Ruflo. Use para diagnosticar ou ajustar a configuracao da equipe. - **test-engineer**: Use this agent when the task involves Design comprehensive unit, integration, contract, and regression test strategies. Examples: - **testing**: Projeta e executa testes, casos extremos e criterios de aceite. Use DEPOIS que o codigo existe e ANTES de considerar a tarefa concluida. - **ui-design-specialist**: Use this agent when the task involves Translate brand direction into consistent components, visual hierarchy, layout, and states. Examples: - **uiux**: Design system, paleta, tipografia, hierarquia visual e fluxo de uso. Use ANTES de implementar interface, para definir a direcao visual. Nao escreve codigo de producao. - **ux-researcher**: Use this agent when the task involves Evaluate user flows, friction, information architecture, and task completion using evidence. Examples: --- ## CAPACIDADE: analytics-instrumentation --- name: analytics-instrumentation description: Add privacy-aware analytics and event instrumentation with stable event contracts. Use when this capability is relevant to the current task. --- # Analytics Instrumentation ## Objective Add privacy-aware analytics and event instrumentation with stable event contracts. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: auditing-dependencies --- name: auditing-dependencies description: Audit package manifests, lockfiles, supply-chain risks, licenses, and drift. Use when this capability is relevant to the current task. --- # Auditing Dependencies ## Objective Audit package manifests, lockfiles, supply-chain risks, licenses, and drift. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: auditing-projects --- name: auditing-projects description: Audit an entire codebase and produce an evidence-based architecture, risk, and readiness report. Use when this capability is relevant to the current task. --- # Auditing Projects ## Objective Audit an entire codebase and produce an evidence-based architecture, risk, and readiness report. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: auditing-secrets --- name: auditing-secrets description: Detect secret exposure patterns without printing secret values. Use when this capability is relevant to the current task. --- # Auditing Secrets ## Objective Detect secret exposure patterns without printing secret values. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: building-memory-systems --- name: building-memory-systems description: Design memory/context strategies with relevance, privacy, compaction, and retrieval safeguards. Use when this capability is relevant to the current task. --- # Building Memory Systems ## Objective Design memory/context strategies with relevance, privacy, compaction, and retrieval safeguards. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: coordinating-agents --- name: coordinating-agents description: Coordinate multiple agents with ownership, conflict avoidance, synthesis, and verification. Use when this capability is relevant to the current task. --- # Coordinating Agents ## Objective Coordinate multiple agents with ownership, conflict avoidance, synthesis, and verification. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: cost-optimization --- name: cost-optimization description: Reduce infrastructure and model costs without sacrificing critical quality. Use when this capability is relevant to the current task. --- # Cost Optimization ## Objective Reduce infrastructure and model costs without sacrificing critical quality. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: debugging-systematically --- name: debugging-systematically description: Diagnose bugs using reproduction, evidence, hypotheses, isolation, fix, and regression checks. Use when this capability is relevant to the current task. --- # Debugging Systematically ## Objective Diagnose bugs using reproduction, evidence, hypotheses, isolation, fix, and regression checks. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: decomposing-tasks --- name: decomposing-tasks description: Decompose large work into independent, agent-ready subtasks with integration boundaries. Use when this capability is relevant to the current task. --- # Decomposing Tasks ## Objective Decompose large work into independent, agent-ready subtasks with integration boundaries. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: deploying-cloudflare --- name: deploying-cloudflare description: Configure and validate Cloudflare Pages/Workers deployments and runtime settings. Use when this capability is relevant to the current task. --- # Deploying Cloudflare ## Objective Configure and validate Cloudflare Pages/Workers deployments and runtime settings. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-agent-workflows --- name: designing-agent-workflows description: Design tool-using agent workflows with loops, stop conditions, verification, and recovery. Use when this capability is relevant to the current task. --- # Designing Agent Workflows ## Objective Design tool-using agent workflows with loops, stop conditions, verification, and recovery. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-apis --- name: designing-apis description: Design API contracts, validation, versioning, pagination, errors, and compatibility. Use when this capability is relevant to the current task. --- # Designing Apis ## Objective Design API contracts, validation, versioning, pagination, errors, and compatibility. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-architecture --- name: designing-architecture description: Create maintainable system architecture with explicit boundaries and trade-offs. Use when this capability is relevant to the current task. --- # Designing Architecture ## Objective Create maintainable system architecture with explicit boundaries and trade-offs. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-auth --- name: designing-auth description: Design secure authentication and authorization flows. Use when this capability is relevant to the current task. --- # Designing Auth ## Objective Design secure authentication and authorization flows. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-databases --- name: designing-databases description: Review data models, constraints, migrations, indexes, and transaction boundaries. Use when this capability is relevant to the current task. --- # Designing Databases ## Objective Review data models, constraints, migrations, indexes, and transaction boundaries. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-design-systems --- name: designing-design-systems description: Create scalable tokens and reusable UI primitives with consistent states. Use when this capability is relevant to the current task. --- # Designing Design Systems ## Objective Create scalable tokens and reusable UI primitives with consistent states. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-mcp --- name: designing-mcp description: Design and troubleshoot MCP server integrations, tool contracts, permissions, and connectivity. Use when this capability is relevant to the current task. --- # Designing Mcp ## Objective Design and troubleshoot MCP server integrations, tool contracts, permissions, and connectivity. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-responsive-ui --- name: designing-responsive-ui description: Design intentional experiences across mobile, tablet, and desktop. Use when this capability is relevant to the current task. --- # Designing Responsive Ui ## Objective Design intentional experiences across mobile, tablet, and desktop. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-ui --- name: designing-ui description: Build cohesive visual systems for typography, color, spacing, components, and states. Use when this capability is relevant to the current task. --- # Designing Ui ## Objective Build cohesive visual systems for typography, color, spacing, components, and states. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-ux --- name: designing-ux description: Optimize user flows, hierarchy, affordances, error recovery, and task completion. Use when this capability is relevant to the current task. --- # Designing Ux ## Objective Optimize user flows, hierarchy, affordances, error recovery, and task completion. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-webhooks --- name: designing-webhooks description: Design reliable webhook handling, idempotency, retries, signatures, and observability. Use when this capability is relevant to the current task. --- # Designing Webhooks ## Objective Design reliable webhook handling, idempotency, retries, signatures, and observability. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: designing-websockets --- name: designing-websockets description: Review real-time channels, reconnect behavior, ordering, backpressure, and failure handling. Use when this capability is relevant to the current task. --- # Designing Websockets ## Objective Review real-time channels, reconnect behavior, ordering, backpressure, and failure handling. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: documenting-apis --- name: documenting-apis description: Generate API reference documentation from observed endpoints and schemas. Use when this capability is relevant to the current task. --- # Documenting Apis ## Objective Generate API reference documentation from observed endpoints and schemas. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-cd --- name: engineering-cd description: Design deployment promotion, verification, rollback, and environment separation. Use when this capability is relevant to the current task. --- # Engineering Cd ## Objective Design deployment promotion, verification, rollback, and environment separation. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-ci --- name: engineering-ci description: Design CI checks and fail-fast gates for build, lint, tests, security, and quality. Use when this capability is relevant to the current task. --- # Engineering Ci ## Objective Design CI checks and fail-fast gates for build, lint, tests, security, and quality. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-content --- name: engineering-content description: Create clear product, brand, and UX copy consistent with the product voice. Use when this capability is relevant to the current task. --- # Engineering Content ## Objective Create clear product, brand, and UX copy consistent with the product voice. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-data-quality --- name: engineering-data-quality description: Detect invalid, duplicated, inconsistent, or stale data and define invariants. Use when this capability is relevant to the current task. --- # Engineering Data Quality ## Objective Detect invalid, duplicated, inconsistent, or stale data and define invariants. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-privacy --- name: engineering-privacy description: Minimize sensitive-data exposure and improve privacy boundaries and retention. Use when this capability is relevant to the current task. --- # Engineering Privacy ## Objective Minimize sensitive-data exposure and improve privacy boundaries and retention. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-prompts --- name: engineering-prompts description: Design robust prompts with explicit goals, constraints, tools, checks, and output formats. Use when this capability is relevant to the current task. --- # Engineering Prompts ## Objective Design robust prompts with explicit goals, constraints, tools, checks, and output formats. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-seo --- name: engineering-seo description: Improve metadata, structured data, indexing, canonicalization, and semantic hierarchy. Use when this capability is relevant to the current task. --- # Engineering Seo ## Objective Improve metadata, structured data, indexing, canonicalization, and semantic hierarchy. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: engineering-tools --- name: engineering-tools description: Design deterministic helper scripts and tools that reduce ambiguity and repeated work. Use when this capability is relevant to the current task. --- # Engineering Tools ## Objective Design deterministic helper scripts and tools that reduce ambiguity and repeated work. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: evaluating-models --- name: evaluating-models description: Evaluate model outputs with task-specific rubrics, adversarial cases, and repeatable tests. Use when this capability is relevant to the current task. --- # Evaluating Models ## Objective Evaluate model outputs with task-specific rubrics, adversarial cases, and repeatable tests. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: feature-flags --- name: feature-flags description: Design safe rollout, targeting, kill switches, and flag lifecycle management. Use when this capability is relevant to the current task. --- # Feature Flags ## Objective Design safe rollout, targeting, kill switches, and flag lifecycle management. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Identify inputs, dependencies, and acceptance criteria. 3. Execute or review the task. 4. Run the strongest available verification. 5. Record facts, remaining risks, and next actions. --- ## CAPACIDADE: final-validation --- name: final-validation description: Run a final multi-discipline readiness review and report only verified facts. Use when this capability is relevant to the current task. --- # Final Validation ## Objective Run a final multi-discipline readiness review and report only verified facts. ## Operating rules - Inspect existing evidence before acting. - Prefer the smallest safe change that satisfies the requirement. - Preserve project conventions and local rules. - Use deterministic tools/scripts for repeatable checks when available. - Never invent execution results. - State blockers precisely. - Validate critical work before declaring it complete. ## Workflow 1. Discover context and constraints. 2. Id …(truncated)