everything-claude-codelisted

# Skill: `everything-claude-code` Catálogo + protocolo de uso da suite ECC. Padrão do projeto: **toda alteração dispara o máximo de agents ECC aplicáveis em paralelo**. Single message, múltiplas tool calls. > **Sempre ativo.** Não opcional. Ignorar = task incompleta. --- ## Trigger - Qualquer edição de código → review agent da stack + security-reviewer (se toca auth/input/segredo). - Build falhou → `*-build-resolver` da stack. - Bug → `engineering:debug` skill + reviewer agent. - Feature nova / refactor amplo → `planner` + `architect` paralelo antes de codar. - E2E → `e2e-runner`. - Docs → `doc-updater` + `docs-lookup` (Context7). - Performance → `performance-optimizer`. - Dead code → `refactor-cleaner`. --- ## Protocolo de invocação (HARD RULE) **Paralelo agressivo é o padrão.** Single message com 3-5+ agents simultâneos sempre que trabalho independente. ### Fluxo por fase | Fase | Agents (paralelo) | |---|---| | Plan (feature/refactor) | `Plan` + `everything-claude-code:planner` + `everything-claude-code:architect` | | Search/Explore | `Explore` (quick/medium/thorough) + `general-purpose` | | Stack review (após edit) | `everything-claude-code:<lang>-reviewer` + `everything-claude-code:code-reviewer` + `everything-claude-code:security-reviewer` | | Build error | `everything-claude-code:<lang>-build-resolver` | | Testes | `everything-claude-code:tdd-guide` + `test-runner` | | E2E | `everything-claude-code:e2e-runner` | | Performance | `everything-claude-code:perform