@orkastery/cli 0.5.6 → 0.6.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/adapters/README.md +18 -0
- package/adapters/claude-code/.claude-plugin/plugin.json +1 -0
- package/adapters/claude-code/README.md +37 -5
- package/adapters/claude-code/agents/ork-check.md +1 -0
- package/adapters/claude-code/agents/ork-go.md +1 -0
- package/adapters/claude-code/agents/ork-goal.md +1 -0
- package/adapters/claude-code/agents/ork-master.md +1 -0
- package/adapters/claude-code/agents/ork-plan.md +1 -0
- package/adapters/claude-code/agents/ork-ship.md +1 -0
- package/adapters/claude-code/hooks/hooks.json +2 -1
- package/adapters/claude-code/hooks/ork-guard.js +15 -1
- package/adapters/claude-code/hooks/ork-resposta.js +91 -0
- package/adapters/claude-code/hooks/ork-sensor.js +18 -2
- package/adapters/claude-code/hooks/quota-signal.js +14 -0
- package/adapters/claude-code/mods/ork-mensageiro/.claude-plugin/plugin.json +14 -0
- package/adapters/claude-code/mods/ork-mensageiro/hooks/caixa.ts +215 -0
- package/adapters/claude-code/mods/ork-mensageiro/hooks/hooks.json +3 -0
- package/adapters/claude-code/mods/ork-mensageiro/hooks/painel.tsx +165 -0
- package/adapters/claude-code/mods/ork-mensageiro/hooks/recepcao.ts +256 -0
- package/adapters/claude-code/mods/ork-mensageiro/hooks/register.tsx +481 -0
- package/adapters/claude-code/mods/ork-mensageiro/orkastery-mod.json +5 -0
- package/adapters/claude-code/mods/ork-mensageiro/types/index.d.ts +85 -0
- package/adapters/claude-code/mods/ork-pista/hooks/ci.ts +9 -5
- package/adapters/claude-code/mods/ork-ritual/.claude-plugin/plugin.json +14 -0
- package/adapters/claude-code/mods/ork-ritual/hooks/hooks.json +3 -0
- package/adapters/claude-code/mods/ork-ritual/hooks/nota.ts +88 -0
- package/adapters/claude-code/mods/ork-ritual/hooks/register.tsx +494 -0
- package/adapters/claude-code/mods/ork-ritual/hooks/replay.tsx +271 -0
- package/adapters/claude-code/mods/ork-ritual/hooks/ritual.ts +218 -0
- package/adapters/claude-code/mods/ork-ritual/orkastery-mod.json +5 -0
- package/adapters/claude-code/mods/ork-ritual/types/index.d.ts +106 -0
- package/adapters/claude-code/mods/ork-sentinela/hooks/vigia.ts +1 -1
- package/adapters/codex/skills/ork/SKILL.md +2 -0
- package/adapters/hermes/README.md +64 -1
- package/adapters/hermes/bin/ork-conhecimento-colecoes.sh +14 -0
- package/adapters/hermes/bin/ork-conhecimento-listar.sh +15 -0
- package/adapters/hermes/bin/ork-conhecimento-mostrar.sh +16 -0
- package/adapters/hermes/bin/ork-conhecimento-verificar.sh +15 -0
- package/adapters/hermes/bin/ork-vigia-batida.py +127 -0
- package/adapters/hermes/hermes.plugin.json +4 -0
- package/adapters/hermes/hitl-ingress/__init__.py +101 -8
- package/adapters/hermes/skills/orkastery-devmaster/SKILL.md +2 -2
- package/adapters/openclaw/README.md +81 -9
- package/adapters/openclaw/bin/ork-vigia-batida.js +86 -0
- package/adapters/openclaw/dist/hitl-ingress.js +106 -9
- package/adapters/openclaw/dist/index.js +86 -0
- package/adapters/openclaw/openclaw.plugin.json +34 -4
- package/adapters/openclaw/package.json +2 -2
- package/adapters/openclaw/src/hitl-ingress.ts +103 -9
- package/adapters/openclaw/src/index.ts +94 -0
- package/assets/adapters-en/claude-code/README.md +403 -0
- package/assets/adapters-en/claude-code/agents/ork-check.md +33 -0
- package/assets/adapters-en/claude-code/agents/ork-go.md +33 -0
- package/assets/adapters-en/claude-code/agents/ork-goal.md +34 -0
- package/assets/adapters-en/claude-code/agents/ork-master.md +35 -0
- package/assets/adapters-en/claude-code/agents/ork-plan.md +33 -0
- package/assets/adapters-en/claude-code/agents/ork-ship.md +33 -0
- package/assets/adapters-en/claude-code/commands/check.md +46 -0
- package/assets/adapters-en/claude-code/commands/go.md +46 -0
- package/assets/adapters-en/claude-code/commands/goal.md +55 -0
- package/assets/adapters-en/claude-code/commands/master.md +54 -0
- package/assets/adapters-en/claude-code/commands/onboarding.md +22 -0
- package/assets/adapters-en/claude-code/commands/ork.md +126 -0
- package/assets/adapters-en/claude-code/commands/plan.md +47 -0
- package/assets/adapters-en/claude-code/commands/ship.md +45 -0
- package/assets/adapters-en/codex/skills/ork/SKILL.md +85 -0
- package/assets/adapters-en/hermes/README.md +273 -0
- package/assets/adapters-en/hermes/skills/orkastery-devmaster/SKILL.md +169 -0
- package/assets/references-en/README.md +25 -0
- package/assets/references-en/code-review-axes.md +82 -0
- package/assets/references-en/definition-of-done.md +74 -0
- package/assets/references-en/performance-checklist.md +53 -0
- package/assets/references-en/security-checklist.md +101 -0
- package/assets/references-en/testing-patterns.md +51 -0
- package/assets/security-scan/defaults/.golangci.yml +2 -0
- package/assets/security-scan/defaults/eslint.config.mjs +11 -0
- package/assets/security-scan/defaults/gitleaks.toml +3 -0
- package/assets/security-scan/defaults/ruff.toml +1 -0
- package/assets/security-scan/tools-agil.json +87 -0
- package/assets/security-scan/tools.json +625 -0
- package/assets/skills-en/core/onboarding/SKILL.md +58 -0
- package/assets/skills-en/core/orkastery-bootstrap/SKILL.md +119 -0
- package/assets/skills-en/core/thread-state/SKILL.md +88 -0
- package/assets/skills-en/governance/decision-triage/SKILL.md +65 -0
- package/assets/skills-en/governance/narrative-guardian/SKILL.md +66 -0
- package/assets/skills-en/governance/roadmap-keeper/SKILL.md +60 -0
- package/assets/skills-en/governance/scope-check-capability-map/SKILL.md +71 -0
- package/assets/skills-en/observability/thread-tracing/SKILL.md +63 -0
- package/assets/skills-en/phases/check-quality/SKILL.md +103 -0
- package/assets/skills-en/phases/go-implementation/SKILL.md +78 -0
- package/assets/skills-en/phases/goal-definition/SKILL.md +69 -0
- package/assets/skills-en/phases/master-metrics/SKILL.md +72 -0
- package/assets/skills-en/phases/plan-specification/SKILL.md +67 -0
- package/assets/skills-en/phases/security-remediation/SKILL.md +70 -0
- package/assets/skills-en/phases/ship-release/SKILL.md +71 -0
- package/assets/skills-en/reviewers/code-reviewer/SKILL.md +61 -0
- package/assets/skills-en/reviewers/security-auditor/SKILL.md +93 -0
- package/assets/skills-en/reviewers/test-engineer/SKILL.md +66 -0
- package/assets/skills-en/reviewers/web-performance-auditor/SKILL.md +63 -0
- package/dist/adapters/claude-bg.js +123 -9
- package/dist/adapters/codex-controller-worker.js +12 -5
- package/dist/adapters/codex-events.js +16 -4
- package/dist/adapters/codex.js +3 -1
- package/dist/adapters/grok-events.js +227 -0
- package/dist/adapters/grok-mcp-servidor.js +612 -0
- package/dist/adapters/grok-mcp.js +178 -0
- package/dist/adapters/grok.js +686 -0
- package/dist/agendador-cpu.js +92 -0
- package/dist/atualizar-branch.js +245 -0
- package/dist/auditoria.js +13 -11
- package/dist/auditrun.js +51 -46
- package/dist/autorizacao-duravel.js +327 -0
- package/dist/board.js +94 -40
- package/dist/branch-de-estado.js +23 -10
- package/dist/busca-semantica.js +7 -5
- package/dist/catalogo.js +2 -1
- package/dist/changelog-reconciliacao.js +290 -0
- package/dist/check-agil.js +528 -0
- package/dist/check-modo.js +41 -0
- package/dist/check-paralelo.js +238 -0
- package/dist/ci.js +248 -45
- package/dist/ciclos.js +8 -6
- package/dist/claim-lint.js +12 -8
- package/dist/claims-conferir.js +47 -39
- package/dist/claims.js +10 -9
- package/dist/company-brain-client.js +10 -3
- package/dist/company-brain-migration.js +11 -5
- package/dist/conducao-batida.js +759 -0
- package/dist/conducao-sentinela.js +25 -8
- package/dist/conducao.js +89 -29
- package/dist/conhecimento-consulta.js +270 -0
- package/dist/conhecimento.js +111 -87
- package/dist/conta-esgotada.js +127 -0
- package/dist/dado-privado.js +816 -0
- package/dist/decisao-autonoma.js +5 -4
- package/dist/demo.js +14 -12
- package/dist/dependencias-da-worktree.js +25 -0
- package/dist/divida.js +38 -36
- package/dist/docs.js +89 -89
- package/dist/doctor-forja.js +343 -0
- package/dist/doctor-pulse-cron.js +113 -1
- package/dist/doctor.js +164 -6
- package/dist/dsn-da-memoria.js +417 -0
- package/dist/entrega-por-mod.js +503 -0
- package/dist/entrega-pr.js +154 -54
- package/dist/esgotamentos.js +13 -11
- package/dist/estado-thread.js +21 -18
- package/dist/evalrunner.js +35 -32
- package/dist/fabrica-batida.js +34 -21
- package/dist/fabrica-estado.js +33 -21
- package/dist/fix.js +21 -19
- package/dist/forja/adaptador.js +6 -0
- package/dist/forja/ambiente-de-ci.js +27 -0
- package/dist/forja/api.js +256 -0
- package/dist/forja/cache.js +89 -0
- package/dist/forja/cli.js +60 -0
- package/dist/forja/contrato.js +9 -0
- package/dist/forja/endereco.js +43 -0
- package/dist/forja/estado.js +22 -0
- package/dist/forja/executor.js +73 -0
- package/dist/forja/github/ci.js +95 -0
- package/dist/forja/github/dono-da-rede.js +18 -0
- package/dist/forja/github/metadados.js +15 -0
- package/dist/forja/github/pedidos.js +200 -0
- package/dist/forja/gitlab/ci.js +141 -0
- package/dist/forja/gitlab/consultas.js +59 -0
- package/dist/forja/gitlab/dono-da-rede.js +43 -0
- package/dist/forja/gitlab/metadados.js +15 -0
- package/dist/forja/gitlab/pedidos-abertos.js +158 -0
- package/dist/forja/gitlab/pedidos.js +141 -0
- package/dist/forja/produtor.js +32 -0
- package/dist/forja/registro.js +31 -0
- package/dist/forja/resolver.js +121 -0
- package/dist/forja/texto.js +13 -0
- package/dist/forja/versao.js +71 -0
- package/dist/forja.js +123 -80
- package/dist/handoff.js +23 -13
- package/dist/hitl-contract.js +46 -43
- package/dist/hitl-curto.js +37 -25
- package/dist/hitl-gates.js +13 -11
- package/dist/hitl-local-receipt.js +51 -0
- package/dist/hitl-local.js +11 -2
- package/dist/hitl-lock.js +5 -11
- package/dist/hitl-public-receipt.js +11 -0
- package/dist/hitl-resumo.js +44 -1
- package/dist/hitl-sessions.js +62 -9
- package/dist/hitl-texto-livre.js +7 -4
- package/dist/hitl.js +34 -31
- package/dist/horario.js +20 -1
- package/dist/hosts.js +104 -134
- package/dist/index.js +864 -554
- package/dist/indice-vetorial.js +14 -12
- package/dist/init.js +11 -2
- package/dist/integracoes-locais.js +4 -0
- package/dist/intelligence-graph-cli.js +123 -69
- package/dist/intelligence-graph-query.js +37 -22
- package/dist/leases.js +19 -18
- package/dist/ledger-stats.js +17 -16
- package/dist/ledger.js +31 -1
- package/dist/licoes.js +12 -11
- package/dist/locale-accounts.js +228 -0
- package/dist/locale-adapter.js +291 -0
- package/dist/locale-ajuda.js +86 -10
- package/dist/locale-audit.js +199 -0
- package/dist/locale-claims.js +213 -0
- package/dist/locale-conducao.js +138 -0
- package/dist/locale-conhecimento.js +181 -0
- package/dist/locale-decisao.js +205 -0
- package/dist/locale-docs.js +230 -0
- package/dist/locale-estado.js +101 -0
- package/dist/locale-forja.js +288 -0
- package/dist/locale-gerados.js +5 -0
- package/dist/locale-grafo.js +129 -0
- package/dist/locale-master.js +83 -0
- package/dist/locale-network.js +524 -0
- package/dist/locale-operacao-restante.js +931 -0
- package/dist/locale-projetos.js +100 -0
- package/dist/locale-prompts.js +194 -0
- package/dist/locale-pulse.js +140 -2
- package/dist/locale-reservas.js +87 -0
- package/dist/locale-retry.js +183 -6
- package/dist/locale-security-scan.js +69 -0
- package/dist/locale-setup.js +192 -0
- package/dist/locale-ship.js +177 -66
- package/dist/locale-thread.js +149 -5
- package/dist/locale-verify.js +80 -4
- package/dist/locale-worktree.js +246 -0
- package/dist/locale.js +553 -9
- package/dist/maestro-runtime.js +22 -3
- package/dist/manifest.js +293 -12
- package/dist/maquina-ociosa.js +211 -0
- package/dist/maquina.js +10 -5
- package/dist/master-nota-local.js +200 -0
- package/dist/master-nota.js +10 -2
- package/dist/master.js +118 -41
- package/dist/mcp-conhecimento.js +52 -0
- package/dist/mcp-experiencia.js +26 -5
- package/dist/mcp-install.js +90 -15
- package/dist/mcp-noctum.js +101 -0
- package/dist/mcp-server.js +116 -22
- package/dist/mcp-ship-transporte.js +977 -0
- package/dist/mcp-ship.js +68 -24
- package/dist/mcp-transportes.js +17 -0
- package/dist/memoria.js +63 -58
- package/dist/merge-da-entrega.js +85 -0
- package/dist/modos-migracao.js +8 -6
- package/dist/modos.js +17 -18
- package/dist/network-roadmap.js +94 -54
- package/dist/noctum-aceite.js +929 -0
- package/dist/noctum-arco.js +540 -33
- package/dist/noctum-capacidade.js +3 -3
- package/dist/noctum-changelog.js +399 -0
- package/dist/noctum-cli.js +539 -0
- package/dist/noctum-coordenacao.js +1225 -0
- package/dist/noctum-entrega.js +1217 -0
- package/dist/noctum-envelope.js +253 -17
- package/dist/noctum-execucao.js +100 -24
- package/dist/noctum-fechamento.js +416 -0
- package/dist/noctum-fila.js +49 -10
- package/dist/noctum-guarda.js +333 -0
- package/dist/noctum-integracao.js +507 -0
- package/dist/noctum-maquinas.js +387 -0
- package/dist/noctum-medicao.js +450 -0
- package/dist/noctum-modo.js +390 -0
- package/dist/noctum-pergunta-canal.js +96 -0
- package/dist/noctum-pergunta.js +115 -0
- package/dist/noctum-provas.js +838 -0
- package/dist/noctum-reconciliacao.js +594 -0
- package/dist/noctum-rede.js +102 -0
- package/dist/noctum-registro.js +215 -0
- package/dist/noctum-relatorio.js +861 -0
- package/dist/noctum-resposta.js +207 -0
- package/dist/noctum-retomada-rede.js +71 -0
- package/dist/noctum-sessoes-em-voo.js +93 -0
- package/dist/noctum-supervisao.js +649 -0
- package/dist/noctum-terminal.js +459 -0
- package/dist/noctum-transporte.js +655 -0
- package/dist/noctum-vigia.js +679 -0
- package/dist/ocupacao.js +3 -1
- package/dist/onboarding.js +16 -5
- package/dist/orkmind.js +39 -23
- package/dist/orquestracao.js +31 -35
- package/dist/parado-no-condutor.js +160 -139
- package/dist/phase.js +96 -42
- package/dist/policies.js +33 -0
- package/dist/postura-local.js +11 -6
- package/dist/preflight.js +2 -1
- package/dist/procedencia.js +9 -14
- package/dist/projeto-alvo.js +59 -20
- package/dist/prompts.js +136 -42
- package/dist/prova-minima.js +5 -6
- package/dist/pulse-cadencia.js +5 -4
- package/dist/pulse-consentimento.js +53 -1
- package/dist/pulse-delivery.js +143 -27
- package/dist/pulse-resposta.js +368 -4
- package/dist/pulse.js +136 -3
- package/dist/ratelimit.js +13 -4
- package/dist/recall.js +14 -13
- package/dist/rede-adesao.js +13 -8
- package/dist/rede-forja.js +108 -24
- package/dist/rede-status.js +63 -45
- package/dist/rede.js +86 -46
- package/dist/retomada-checkpoint.js +101 -0
- package/dist/retry.js +326 -78
- package/dist/roadmap-reservas.js +32 -32
- package/dist/runtime-ambiente.js +5 -0
- package/dist/runtime-context.js +279 -85
- package/dist/runtime-escada.js +429 -0
- package/dist/runtime-profiles.js +130 -63
- package/dist/runtimes.js +42 -5
- package/dist/saida-segura.js +3 -2
- package/dist/security-scan/agil.js +366 -0
- package/dist/security-scan/cli.js +76 -0
- package/dist/security-scan/compare.js +149 -0
- package/dist/security-scan/config.js +93 -0
- package/dist/security-scan/detect.js +97 -0
- package/dist/security-scan/executor.js +196 -0
- package/dist/security-scan/guard.js +180 -0
- package/dist/security-scan/imagem-de-lint.js +96 -0
- package/dist/security-scan/normalize/actionlint.js +12 -0
- package/dist/security-scan/normalize/eslint.js +12 -0
- package/dist/security-scan/normalize/gitleaks.js +13 -0
- package/dist/security-scan/normalize/golangci-lint.js +13 -0
- package/dist/security-scan/normalize/govulncheck.js +36 -0
- package/dist/security-scan/normalize/index.js +25 -0
- package/dist/security-scan/normalize/npm-audit.js +30 -0
- package/dist/security-scan/normalize/osv-scanner.js +32 -0
- package/dist/security-scan/normalize/oxlint.js +51 -0
- package/dist/security-scan/normalize/pip-audit.js +16 -0
- package/dist/security-scan/normalize/ruff.js +12 -0
- package/dist/security-scan/normalize/semgrep.js +20 -0
- package/dist/security-scan/normalize/trivy.js +33 -0
- package/dist/security-scan/normalize/util.js +204 -0
- package/dist/security-scan/normalize/zizmor.js +19 -0
- package/dist/security-scan/report.js +156 -0
- package/dist/security-scan/run.js +448 -0
- package/dist/security-scan/tools.js +62 -0
- package/dist/security-scan/trabalhador.js +74 -0
- package/dist/security-scan/types.js +15 -0
- package/dist/session-events.js +19 -2
- package/dist/session-watcher-claude.js +74 -19
- package/dist/session-watcher-grok.js +425 -0
- package/dist/session-watcher.js +174 -13
- package/dist/sessoes-contas.js +42 -5
- package/dist/sessoes-inventario.js +1 -1
- package/dist/sessoes.js +25 -23
- package/dist/setup.js +67 -66
- package/dist/ship-publicar.js +263 -92
- package/dist/ship.js +160 -17
- package/dist/superficie.js +14 -11
- package/dist/thread-close.js +9 -9
- package/dist/thread.js +121 -18
- package/dist/tokens.js +15 -11
- package/dist/trabalho-thread.js +297 -0
- package/dist/transporte-do-canal.js +189 -0
- package/dist/types.js +12 -1
- package/dist/util.js +5 -2
- package/dist/varredura-acoes.js +500 -0
- package/dist/verify.js +491 -32
- package/dist/worktree-local.js +12 -10
- package/dist/worktree.js +81 -85
- package/eval/casos/check-quality.json +77 -0
- package/eval/casos/security-auditor.json +148 -0
- package/eval/casos/security-remediation.json +161 -0
- package/eval/fixtures/security-scan/README.md +8 -0
- package/eval/fixtures/security-scan/go/.github/workflows/ci.yml +12 -0
- package/eval/fixtures/security-scan/go/Dockerfile +4 -0
- package/eval/fixtures/security-scan/go/README.md +4 -0
- package/eval/fixtures/security-scan/go/config.go +4 -0
- package/eval/fixtures/security-scan/go/go.mod +5 -0
- package/eval/fixtures/security-scan/go/go.sum +2 -0
- package/eval/fixtures/security-scan/go/main.go +31 -0
- package/eval/fixtures/security-scan/js/.github/workflows/ci.yml +12 -0
- package/eval/fixtures/security-scan/js/Dockerfile +3 -0
- package/eval/fixtures/security-scan/js/README.md +4 -0
- package/eval/fixtures/security-scan/js/package-lock.json +844 -0
- package/eval/fixtures/security-scan/js/package.json +10 -0
- package/eval/fixtures/security-scan/js/src/config.js +4 -0
- package/eval/fixtures/security-scan/js/src/server.js +17 -0
- package/eval/fixtures/security-scan/js/src/util.ts +4 -0
- package/eval/fixtures/security-scan/js/tsconfig.json +4 -0
- package/eval/fixtures/security-scan/python/.github/workflows/ci.yml +12 -0
- package/eval/fixtures/security-scan/python/Dockerfile +3 -0
- package/eval/fixtures/security-scan/python/README.md +4 -0
- package/eval/fixtures/security-scan/python/app/__init__.py +0 -0
- package/eval/fixtures/security-scan/python/app/config.py +2 -0
- package/eval/fixtures/security-scan/python/app/main.py +19 -0
- package/eval/fixtures/security-scan/python/requirements.txt +1 -0
- package/package.json +10 -5
- package/references/security-checklist.md +3 -3
- package/skills/README.md +1 -0
- package/skills/core/orkastery-bootstrap/SKILL.md +2 -2
- package/skills/phases/check-quality/SKILL.md +16 -0
- package/skills/phases/go-implementation/SKILL.md +6 -0
- package/skills/phases/security-remediation/SKILL.md +70 -0
- package/skills/reviewers/security-auditor/SKILL.md +25 -0
|
@@ -0,0 +1,46 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Conducts the thread's CHECK phase and consolidates a single verdict.
|
|
3
|
+
argument-hint: "<thread-id>"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /check
|
|
8
|
+
|
|
9
|
+
CHECK phase (F4): verification against the baseline, the five review axes, security with zero blockers, a delegation audit and exactly one verdict.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Re-runs everything on the real HEAD:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
ork verify <thread>
|
|
17
|
+
ork worktree audit <thread>
|
|
18
|
+
```
|
|
19
|
+
|
|
20
|
+
2. Runs the four reviewers from the catalog over the declared scope: `code-reviewer`,
|
|
21
|
+
`security-auditor`, `test-engineer` and `web-performance-auditor`. Use the
|
|
22
|
+
`ork-check` subagent and its siblings when you want to isolate each one's context.
|
|
23
|
+
|
|
24
|
+
3. Reads the ledger for the delegation audit:
|
|
25
|
+
|
|
26
|
+
```bash
|
|
27
|
+
ork phase list <thread>
|
|
28
|
+
```
|
|
29
|
+
|
|
30
|
+
4. Consolidates ONE verdict (PASSOU = passed, PRECISA DE MUDANCA = needs changes, BLOQUEADO = blocked) and, if the mode pauses, opens the request
|
|
31
|
+
and waits. The builder answers through the authenticated channel (the host's native dialog, or
|
|
32
|
+
`/ork gate <thread> <pedido> <resposta>` in Telegram); the agent never answers for the builder:
|
|
33
|
+
|
|
34
|
+
```bash
|
|
35
|
+
ork gate request <thread>
|
|
36
|
+
```
|
|
37
|
+
|
|
38
|
+
## Adapter rules
|
|
39
|
+
|
|
40
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
41
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
42
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
43
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
44
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
45
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
46
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,46 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Conducts the thread's GO phase through `ork`, inside the isolated worktree.
|
|
3
|
+
argument-hint: "<thread-id> [task]"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /go
|
|
8
|
+
|
|
9
|
+
GO phase (F3): implementation slice by slice, one atomic commit per task, inside the thread's worktree. Code is written by the dispatched session, never by this command.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Ensures the worktree and records the baseline BEFORE the first line of code:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
ork worktree ensure <thread>
|
|
17
|
+
ork verify <thread> --baseline
|
|
18
|
+
```
|
|
19
|
+
|
|
20
|
+
2. Dispatches the task:
|
|
21
|
+
|
|
22
|
+
```bash
|
|
23
|
+
ork phase run <thread> GO --prompt "<tarefa do PLAN, uma por vez>"
|
|
24
|
+
```
|
|
25
|
+
|
|
26
|
+
3. Re-runs the verification on the real HEAD and compares it with the baseline:
|
|
27
|
+
|
|
28
|
+
```bash
|
|
29
|
+
ork verify <thread>
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
4. Checks that the worktree stays consistent in git itself:
|
|
33
|
+
|
|
34
|
+
```bash
|
|
35
|
+
ork worktree audit <thread>
|
|
36
|
+
```
|
|
37
|
+
|
|
38
|
+
## Adapter rules
|
|
39
|
+
|
|
40
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
41
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
42
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
43
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
44
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
45
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
46
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Opens the thread for the request and conducts the GOAL phase through `ork`.
|
|
3
|
+
argument-hint: "<builder request, with optional #TAG>"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /goal
|
|
8
|
+
|
|
9
|
+
GOAL phase (F1): a verifiable goal, exploration of the real repository, explicit assumptions and criteria that become claims with a command. GOAL does not implement.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Extracts the conduction mode from the request itself, without reimplementing the parsing:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
MODO=$(ork modos --do-pedido "$ARGUMENTS")
|
|
17
|
+
```
|
|
18
|
+
|
|
19
|
+
2. Opens the thread with that mode (`ork` validates against `conduction.allowed_modes` and refuses what
|
|
20
|
+
the project does not allow):
|
|
21
|
+
|
|
22
|
+
```bash
|
|
23
|
+
ork thread new "<nome curto da demanda>" --mode "$MODO" --worktree auto
|
|
24
|
+
```
|
|
25
|
+
|
|
26
|
+
3. Dispatches the phase, with the builder's full request:
|
|
27
|
+
|
|
28
|
+
```bash
|
|
29
|
+
ork phase run <thread> GOAL --prompt "$ARGUMENTS"
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
4. Records as a claim everything GOAL stated, with the command that proves each statement, and
|
|
33
|
+
re-runs:
|
|
34
|
+
|
|
35
|
+
```bash
|
|
36
|
+
ork claims add <thread> <arquivo> --claim "<alegacao>" --verificar "<comando>" --fase GOAL
|
|
37
|
+
ork verify <thread> --so-claims
|
|
38
|
+
```
|
|
39
|
+
|
|
40
|
+
5. If the mode pauses in this block, presents the evidence to the builder and stops. The request comes from the core, and the
|
|
41
|
+
release is only recorded when the builder answers through the authenticated channel:
|
|
42
|
+
|
|
43
|
+
```bash
|
|
44
|
+
ork gate request <thread>
|
|
45
|
+
```
|
|
46
|
+
|
|
47
|
+
## Adapter rules
|
|
48
|
+
|
|
49
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
50
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
51
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
52
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
53
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
54
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
55
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,54 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Closes the thread with a typed POSTMORTEM and the human score from 0 to 5.
|
|
3
|
+
argument-hint: "<thread-id>"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /master
|
|
8
|
+
|
|
9
|
+
MASTER phase (F6): typed POSTMORTEM, MASTER log in the frozen contract and the human score with a justification. A delivery without a MASTER log did not happen.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Shows the builder the failure classes from the fixed catalog:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
ork master classes
|
|
17
|
+
```
|
|
18
|
+
|
|
19
|
+
2. Asks the owner for the score through the channel that carries proof (RM-048). **The score belongs to the human, in every mode.**
|
|
20
|
+
This command never makes up a score and never writes `--por` on anyone's behalf:
|
|
21
|
+
|
|
22
|
+
```bash
|
|
23
|
+
ork master pedir <thread> --formato terminal
|
|
24
|
+
```
|
|
25
|
+
|
|
26
|
+
Show the line as it comes. The owner replies through Telegram (`<codigo> <0 a 5> <porque>`) and the core
|
|
27
|
+
records the score with the ingress receipt. `ork master <thread> --score ... --por` called from here is
|
|
28
|
+
refused with `master.prova-de-canal`; from their own terminal, outside the host, the owner can use it.
|
|
29
|
+
3. Check that the score made it into the ledger:
|
|
30
|
+
|
|
31
|
+
```bash
|
|
32
|
+
ork master --todas
|
|
33
|
+
```
|
|
34
|
+
|
|
35
|
+
4. In modes with no MASTER pause, the delivery is accepted unless the builder says otherwise;
|
|
36
|
+
`ork master --todas` shows the deliveries with the ledger-derived index, including those already scored.
|
|
37
|
+
To close by default, always pass the thread, so that only yours is closed:
|
|
38
|
+
|
|
39
|
+
```bash
|
|
40
|
+
ork master <thread> --aceitar-omissao
|
|
41
|
+
```
|
|
42
|
+
|
|
43
|
+
Without the thread, the command accepts ALL of the project's delivered threads, including those from other parallel
|
|
44
|
+
workstreams; `ork master --aceitar-omissao --dry-run` shows which ones, without recording.
|
|
45
|
+
|
|
46
|
+
## Adapter rules
|
|
47
|
+
|
|
48
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
49
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
50
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
51
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
52
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
53
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
54
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,22 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Conducts or resumes the project interview following the agenda from the ork core.
|
|
3
|
+
argument-hint: "<onboarding request>"
|
|
4
|
+
allowed-tools: Bash(ork:*)
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /onboarding
|
|
8
|
+
|
|
9
|
+
Run `ork onboarding` to get the agenda and `ork onboarding show --json` to look up
|
|
10
|
+
the current answers. Work through the pending items following the agenda returned by the core. Do not duplicate
|
|
11
|
+
steps or validation in this host, and do not make up answers on the builder's behalf.
|
|
12
|
+
|
|
13
|
+
Record the requested public answer with `ork onboarding set <etapa> --conteudo <JSON> --por <quem>`.
|
|
14
|
+
Pass the JSON as an argument, with proper shell escaping; never execute the text you received.
|
|
15
|
+
Ask only for variable names for credentials. Secret values stay in the process environment
|
|
16
|
+
or in the host's vault (in Hermes, `~/.hermes/.env`); do not read them or carry values into the
|
|
17
|
+
conversation, the onboarding or the ledger.
|
|
18
|
+
|
|
19
|
+
A requested reset uses `ork onboarding reset [etapa]`. Optional publication uses
|
|
20
|
+
`ork onboarding sync --json`; present the typed degradation reason without blocking the interview.
|
|
21
|
+
The memory choice guides the manifest, without editing it automatically.
|
|
22
|
+
Finish with the state from `ork onboarding show --json`. Do not create a thread just to interview.
|
|
@@ -0,0 +1,126 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Recognizes orkastery maestro and conducts requests, resumptions and status in the conversation with the Maestro/Builder.
|
|
3
|
+
argument-hint: "[natural-language request|thread|status|board|plan|doctor|modos]"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, mcp__orkastery__ork_thread_status, mcp__orkastery__ork_phase_list, mcp__orkastery__ork_hitl_pending, mcp__orkastery__ork_observe, mcp__orkastery__ork_artifact_read, mcp__orkastery__ork_claims_list, mcp__orkastery__ork_git_status, mcp__orkastery__ork_thread_new, mcp__orkastery__ork_phase_run, mcp__orkastery__ork_preflight
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /orkastery:ork
|
|
8
|
+
|
|
9
|
+
Native entry point of the plugin enabled in the project: `/orkastery:ork`. Copying the files through the
|
|
10
|
+
installer still requires native activation; `/ork` alone is not a guaranteed alias.
|
|
11
|
+
|
|
12
|
+
This entry point is already loaded by the owner's invocation. Do not invoke `Skill(orkastery:ork)`
|
|
13
|
+
or another command to reload these instructions. Conduct the request with the tools
|
|
14
|
+
already allowed.
|
|
15
|
+
The eight MCP queries and creating/dispatching a thread from this entry point are valid for the invocation turn;
|
|
16
|
+
they do not persist on their own into the next messages. Creating/dispatching requires a demand and scope authorized
|
|
17
|
+
by the owner, a resolved mode and the core's preflight. Do not reload the skill to obtain permissions.
|
|
18
|
+
These grants do not enable MCP, do not authorize other mutations and do not approve gates or human answers.
|
|
19
|
+
|
|
20
|
+
Read the common contract with `Read`, without calling `Skill` to load the bootstrap:
|
|
21
|
+
@${CLAUDE_PLUGIN_ROOT}/skills/core/orkastery-bootstrap/SKILL.md
|
|
22
|
+
|
|
23
|
+
If the contract is already in context, use it without rereading. If the file is missing,
|
|
24
|
+
report an incomplete installation; do not invent another flow.
|
|
25
|
+
|
|
26
|
+
Orkastery's conversational entry point. Use the common contract to conduct the owner's intent through the core. Do not limit an execution request to pointing to another
|
|
27
|
+
slash command. If you received a request and have the necessary tools, carry out the routing.
|
|
28
|
+
The phase commands remain available as shortcuts; the owner does not need to know them.
|
|
29
|
+
|
|
30
|
+
## Recognize the intent
|
|
31
|
+
|
|
32
|
+
The literal phrase `orkastery maestro`, in a clean session, asks for a read of the overview.
|
|
33
|
+
Discover the tool with `ToolSearch` on `select:mcp__orkastery__ork_maestro`,
|
|
34
|
+
then query the pinned project. A query does not open a thread or a dispatch.
|
|
35
|
+
With no project named, also offer the network overview (RM-054): `ork network roadmap`
|
|
36
|
+
(all known projects, the threads of all machines, sources, freshness and gaps)
|
|
37
|
+
or `mcp__orkastery__ork_network_roadmap` (the pinned project, on all machines), shown as it comes.
|
|
38
|
+
The person's machines (RM-053) come from `ork network status` or from `mcp__orkastery__ork_network_status`
|
|
39
|
+
(in MCP, only the pinned project on each machine), shown as it comes: a gap is never "no machines".
|
|
40
|
+
MCP serves only the pinned project; `projeto` in any tool only checks that
|
|
41
|
+
project (`projeto.fora-do-servidor` when it is a different one). A request for another project goes to the CLI with
|
|
42
|
+
`--projeto <nome>` (RM-052). Read "Projeto consultado" (project queried) and "Não lido" (not read) before answering.
|
|
43
|
+
A missing/ambiguous project requires resolving the context allowed by the core.
|
|
44
|
+
If you already received a phase/thread/worktree, execute the received block, without recursion.
|
|
45
|
+
Missing MCP or denied authentication remains a blocker; do not switch runtime,
|
|
46
|
+
provider or sandbox. The CLI query `ork maestro --json` does not grant mutations.
|
|
47
|
+
|
|
48
|
+
HITL usability is top priority: bullets, a recommendation and labeled options,
|
|
49
|
+
internal UUID and preserved cancellation. The conductor session uses
|
|
50
|
+
`mcp__orkastery__ork_request_decision`; the child session never answers on the owner's behalf.
|
|
51
|
+
|
|
52
|
+
Conduction HITL is a selection (RM-057): 3 to 5 alternatives, exactly one with the
|
|
53
|
+
"Recommendation" seal. Never ask the owner for an "I confirm" in free text or to paste text; the request
|
|
54
|
+
they pasted with explicit authorization counts as their instruction, within that authorization.
|
|
55
|
+
A doubt within the authorization becomes `ork decisao registrar` and the thread goes on. Text only when the
|
|
56
|
+
factory cannot go on by itself and the owner needs to run a command in the terminal: show the
|
|
57
|
+
exact command.
|
|
58
|
+
|
|
59
|
+
| Owner request | Agent action |
|
|
60
|
+
|---|---|
|
|
61
|
+
| New demand | Resolve the mode through the core, check the preflight and open/dispatch the corresponding thread. |
|
|
62
|
+
| "Continue" or an existing thread/roadmap | Read the state and the last result; resume the next authorized step, without duplicating the dispatch. |
|
|
63
|
+
| "How is it going?" or status | Query the known thread; present proven progress, the blocker and the next step. |
|
|
64
|
+
| Roadmap status | Run `ork network roadmap` (with `--projeto <nome>` when the owner names the project) and show the text as it comes: it brings the threads of all machines, the reservations and the source and time of each part; `ork roadmap status` reads only this machine. Do not write your own roadmap report or deduce it from the board or the overview: zero threads is not an empty roadmap, and a gap or "Nao consultado" (not queried) is never an empty roadmap. |
|
|
65
|
+
| Decision or answer | Correlate it with the presented request/gate, relay it through the supported path and check the result. |
|
|
66
|
+
| Onboarding | Use the agenda of `ork onboarding` and resume the existing steps. |
|
|
67
|
+
| board, plan, doctor or modos | Run the corresponding query and interpret the result for the owner's intent. |
|
|
68
|
+
|
|
69
|
+
## Core sources
|
|
70
|
+
|
|
71
|
+
Prioritize Orkastery's MCP tools when discovered in this session, with the project
|
|
72
|
+
pinned by the server. In Claude, the installed `orkastery` server exposes full names such as
|
|
73
|
+
`mcp__orkastery__ork_thread_status`. If there are deferred tools, use `ToolSearch` with
|
|
74
|
+
`select:mcp__orkastery__ork_thread_status,mcp__orkastery__ork_artifact_read` for those reads;
|
|
75
|
+
for another operation, select its full name as actually exposed. Do not try `select:ork_thread_status`
|
|
76
|
+
or guess another namespace. The initial MCP activation and the owner's consent remain native. `ork_thread_status`, `ork_phase_list` and `ork_hitl_pending` read
|
|
77
|
+
state; `ork_thread_new` and `ork_phase_run` open/dispatch; `ork_observe` queries the dispatched
|
|
78
|
+
native session. When a real pause/escalation allows it and a request is missing, use
|
|
79
|
+
`ork_gate_request(threadId, motivo?)`; never create a request for a status query.
|
|
80
|
+
`ork_request_decision(threadId, pedidoId)` presents the dialog to the owner:
|
|
81
|
+
the answer comes through the host, never through an answer argument or an identity invented by the model.
|
|
82
|
+
|
|
83
|
+
The CLI queries below remain useful when the corresponding tool is not available:
|
|
84
|
+
|
|
85
|
+
```bash
|
|
86
|
+
ork thread status <thread>
|
|
87
|
+
ork phase list <thread>
|
|
88
|
+
ork network roadmap [--projeto <nome>]
|
|
89
|
+
ork pulse --json
|
|
90
|
+
ork board plan
|
|
91
|
+
ork doctor
|
|
92
|
+
ork modos
|
|
93
|
+
ork master
|
|
94
|
+
```
|
|
95
|
+
|
|
96
|
+
Use only the necessary queries. Do not run all of them on every message. The preflight belongs to the
|
|
97
|
+
start of a dispatch or to a relevant environment change; a status question does not restart
|
|
98
|
+
onboarding or work. If the thread is ambiguous, clarify before changing its state.
|
|
99
|
+
Pauses listed in the mode's blocks are expected. To say that the owner needs to answer
|
|
100
|
+
now, identify a current pending item in the thread's state/last result, with a matching gate,
|
|
101
|
+
request or event. A new thread with no phase executed is not waiting on assumptions
|
|
102
|
+
just because the mode's f12 block provides for that pause.
|
|
103
|
+
|
|
104
|
+
## Continuity
|
|
105
|
+
|
|
106
|
+
Preserve the thread/session mode, decisions and authorizations. Proceed with authorized work instead
|
|
107
|
+
of ending with "may I continue?". Ask when the mode or a real ambiguity requires
|
|
108
|
+
the owner's intervention, and explain what the answer unlocks. Do not approve gates on their behalf.
|
|
109
|
+
|
|
110
|
+
With an existing dispatch, query `ork_observe` and handle the current native wait. Do not replace
|
|
111
|
+
observation with long loops of `phase list`, sleeps or completion regex over history.
|
|
112
|
+
If MCP is missing and the sandbox blocks `.git`, report that the native transport installation is pending;
|
|
113
|
+
do not repeat the denied mutation or change the sandbox. Tell a native permission apart from an Ork gate.
|
|
114
|
+
|
|
115
|
+
Before the dispatch, use `ork_preflight({modo})`; CLI alternative: `ork doctor --modo <modo>`. The probe follows the per-block setup, without a global inventory, and does not prove authentication, gates or delivery.
|
|
116
|
+
For HEADs and Git state, query `ork_git_status`; do not invent SHAs or replace the transport with Bash Git.
|
|
117
|
+
The `worktree` child profile is an opt-in of the project's installation; do not enable it through a tool argument
|
|
118
|
+
or a configuration edit to bypass a block. Preserve the configured profile and the native permissions.
|
|
119
|
+
|
|
120
|
+
The conductor session dispatches and follows up; a session that already received a phase executes that phase,
|
|
121
|
+
without redispatching itself. The core keeps validating phases, evidence and gates.
|
|
122
|
+
|
|
123
|
+
Show the result, the last confirmed progress and what happens now. Record a checkpoint
|
|
124
|
+
before a future resumption; do not promise notification beyond the session's capabilities.
|
|
125
|
+
Conclude with effective delivery and validation, distinguishing what still depends on CHECK, SHIP or
|
|
126
|
+
human score. The list of executed commands is auxiliary diagnostics, not the owner's experience.
|
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Conducts the thread's PLAN phase through `ork`.
|
|
3
|
+
argument-hint: "<thread-id> [note]"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /plan
|
|
8
|
+
|
|
9
|
+
PLAN phase (F2): sliced tasks with `touch_paths` and a verify per task, decisions D1..Dn with a single home. PLAN does not implement.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Reads the real state of the thread before anything else:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
ork thread status <thread>
|
|
17
|
+
```
|
|
18
|
+
|
|
19
|
+
2. Dispatches the phase:
|
|
20
|
+
|
|
21
|
+
```bash
|
|
22
|
+
ork phase run <thread> PLAN --prompt "<o que o plano precisa cobrir>"
|
|
23
|
+
```
|
|
24
|
+
|
|
25
|
+
3. Reserves the regions the plan declared it would touch, so that another thread does not write over them:
|
|
26
|
+
|
|
27
|
+
```bash
|
|
28
|
+
ork lease acquire "path:<glob>" --thread <thread> --motivo "PLAN reservou a regiao"
|
|
29
|
+
```
|
|
30
|
+
|
|
31
|
+
4. Exports the typed handoff to the next phase, instead of relying on the session's memory:
|
|
32
|
+
|
|
33
|
+
```bash
|
|
34
|
+
ork handoff export <thread> --proxima-fase GO
|
|
35
|
+
```
|
|
36
|
+
|
|
37
|
+
5. If the mode pauses, presents the tradeoffs with options and a recommendation, and waits for the verdict.
|
|
38
|
+
|
|
39
|
+
## Adapter rules
|
|
40
|
+
|
|
41
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
42
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
43
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
44
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
45
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
46
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
47
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,45 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: "Delivers the thread through `ork ship`: serialized merge and proven push."
|
|
3
|
+
argument-hint: "<thread-id> [--para <branch>]"
|
|
4
|
+
allowed-tools: Bash(ork:*), Read, Glob, Grep
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# /ship
|
|
8
|
+
|
|
9
|
+
SHIP phase (F5): merge serialized by lease, update against the base, push proven by command and a rollback plan. No `git push` by hand: a push by hand is not proven, and the per-project adapter guard (`ork adapter install claude-code`) blocks it.
|
|
10
|
+
|
|
11
|
+
## What this command does
|
|
12
|
+
|
|
13
|
+
1. Resyncs when the base has moved:
|
|
14
|
+
|
|
15
|
+
```bash
|
|
16
|
+
ork worktree sync <thread>
|
|
17
|
+
```
|
|
18
|
+
|
|
19
|
+
2. Rehearses the whole delivery before touching anything:
|
|
20
|
+
|
|
21
|
+
```bash
|
|
22
|
+
ork ship <thread> --para main --dry-run
|
|
23
|
+
```
|
|
24
|
+
|
|
25
|
+
3. Delivers, with the authorization recorded:
|
|
26
|
+
|
|
27
|
+
```bash
|
|
28
|
+
ork ship <thread> --para main --autorizar-push "<quem>"
|
|
29
|
+
```
|
|
30
|
+
|
|
31
|
+
4. Shows the proof to the builder: `mergeSha`, `shaRemoto` and `pushVerificado` in the ledger.
|
|
32
|
+
|
|
33
|
+
```bash
|
|
34
|
+
ork phase list <thread>
|
|
35
|
+
```
|
|
36
|
+
|
|
37
|
+
## Adapter rules
|
|
38
|
+
|
|
39
|
+
- **Zero business logic here.** This command turns intent into an `ork` call and nothing more.
|
|
40
|
+
Validating the mode, building the prompt, deciding the gate and proving the delivery is the core's job.
|
|
41
|
+
- **The #TAG comes from the request, not from configuration.** The mode comes from `ork modos --do-pedido`, which uses the
|
|
42
|
+
same function as the core; without a tag, the manifest's `conduction.default_mode` applies. Validation
|
|
43
|
+
against `conduction.allowed_modes` happens inside `ork thread new`, never here.
|
|
44
|
+
- **No writing product code in this command.** Writing is the job of the session that `ork`
|
|
45
|
+
dispatches, with the prompt recorded and its sha in the ledger.
|
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: ork
|
|
3
|
+
description: "Recognizes orkastery maestro in a clean session: queries the overview, resumes threads and presents native decisions in the Codex conversation with the Maestro/Builder."
|
|
4
|
+
license: MIT
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
# Orkastery in Codex
|
|
8
|
+
|
|
9
|
+
When you receive `orkastery maestro` literally, the first action is the `ork_maestro` tool of the
|
|
10
|
+
`orkastery` MCP server, discovered in the namespace actually exposed: it queries the pinned
|
|
11
|
+
project. With the tool exposed, do not run `ork maestro`, `ork doctor`, `ork onboarding` or
|
|
12
|
+
`ork experiencia show` in the shell before it or in its place: the CLI resolves the project by the session directory, and that is the
|
|
13
|
+
class of error the phrase avoids (Codex activation proof, RM-032).
|
|
14
|
+
|
|
15
|
+
Only after `ork_maestro` (or, without MCP, `ork maestro --json`), query `ork experiencia show --json`
|
|
16
|
+
in the session's project. When `experience` is true,
|
|
17
|
+
read `../core/<skill>/SKILL.md`, using the name returned in `skill` by the core:
|
|
18
|
+
`orchestration-experience-pt-br` or `orchestration-experience`. Language, timezone and depth
|
|
19
|
+
come from the same query. If false, do not activate the pack. The dedicated block in AGENTS.md also
|
|
20
|
+
points to this catalog; installation does not change native permissions.
|
|
21
|
+
|
|
22
|
+
After `ork_maestro`, the phrase `orkastery maestro` still asks only for the overview,
|
|
23
|
+
without opening work. With no project named, also offer the network overview
|
|
24
|
+
(`ork_network_roadmap`, or `ork network roadmap` in the CLI), with sources, freshness and gaps.
|
|
25
|
+
Follow the context selection and the limits of the common bootstrap.
|
|
26
|
+
Without MCP, the `ork maestro --json` query stays read-only if the CLI is
|
|
27
|
+
available; mutations without native transport stay pending. Never use Bash Git
|
|
28
|
+
or a sandbox change as substitutes. A child session continues the received phase.
|
|
29
|
+
|
|
30
|
+
HITL is top priority in the presentation: short bullets, a recommendation and options
|
|
31
|
+
with clear labels; internal UUID. Use `ork_request_decision` in the conductor session and preserve
|
|
32
|
+
the literal answer/cancellation of the elicitation. Do not use a generic agent question
|
|
33
|
+
as gate approval, nor text in toolargs as human identity.
|
|
34
|
+
|
|
35
|
+
Conduction HITL is a selection (RM-057): 3 to 5 alternatives, exactly one with the
|
|
36
|
+
"Recommendation" seal. Never ask the owner for an "I confirm" in free text or to paste text; the request
|
|
37
|
+
they pasted with explicit authorization counts as their instruction, within that authorization.
|
|
38
|
+
A doubt within the authorization becomes `ork decisao registrar` and the thread goes on. Text only when the
|
|
39
|
+
factory cannot go on by itself and the owner needs to run a command in the terminal: show the
|
|
40
|
+
exact command.
|
|
41
|
+
|
|
42
|
+
Use the `orkastery-bootstrap` skill from the same installed catalog to conduct the request.
|
|
43
|
+
The owner talks about the goal; you operate the `ork` core and present the result and the next
|
|
44
|
+
step. Preserve the thread's mode and authorizations. If you already received an Orkastery phase,
|
|
45
|
+
run that phase within the received scope, without redispatching yourself.
|
|
46
|
+
|
|
47
|
+
- Waits: never start verify, build or the suite with `&` inside the command, because the process gets lost and the session waits for a notice that never comes; run `ork verify` in the foreground, with `timeout` (it waits its turn in the heavy-resource queue), or use the host's background execution, which tells you when it finishes; never wait with `pgrep -f` whose pattern shows up in the command itself, because the loop finds itself and never ends: wait for the output file with a deadline, or for the saved PID.
|
|
48
|
+
|
|
49
|
+
When discovered, prefer the MCP tools bound to the project by the server:
|
|
50
|
+
`ork_thread_status`, `ork_phase_list`, `ork_hitl_pending`, `ork_thread_new(nome, modo)`,
|
|
51
|
+
`ork_phase_run(threadId, fase, prompt, runtime?, model?, effort?, dryRun?)`, `ork_observe(threadId)`
|
|
52
|
+
and `ork_request_decision(threadId, pedidoId)`. Use the namespace actually exposed, without assuming another project.
|
|
53
|
+
|
|
54
|
+
- New request: resolve the mode through `ork`, check preflight and open/dispatch the thread.
|
|
55
|
+
- Resumption: read `ork thread status <thread>` and the last result in `ork phase list <thread>`;
|
|
56
|
+
continue the authorized work without repeating onboarding, approval or implementation.
|
|
57
|
+
- Roadmap status: use `ork_network_roadmap` (or `ork network roadmap --projeto <nome>`) and show the
|
|
58
|
+
text as it comes, with the threads of all machines, the sources and the gaps; `ork_roadmap_status` is
|
|
59
|
+
only for this machine. Do not write your own roadmap report: a gap is never an empty roadmap.
|
|
60
|
+
- The person's machines (RM-053): use `ork_network_status` (or `ork network status`) and show the text
|
|
61
|
+
as it comes, with the beat of each machine, the source and the gaps; a gap is never "no machines".
|
|
62
|
+
- Status: present the thread's evidence; `ork pulse --json` shows who needs to act.
|
|
63
|
+
The pauses of the mode's blocks are expected: only say "waiting for you" with a current human
|
|
64
|
+
dependency proven in the state/last result, identifying the gate, request or event.
|
|
65
|
+
A new thread with no phase run and no open request is not yet awaiting a verdict.
|
|
66
|
+
- Existing dispatch: check `ork_observe` for the current native progress or block;
|
|
67
|
+
do not wait for completion with long `phase list` loops, sleeps or regex over history.
|
|
68
|
+
- Decision: if a real pause/escalation allows it and a request is missing, use `ork_gate_request(threadId, motivo?)`;
|
|
69
|
+
never create a request for status. Correlate thread/request and use `ork_request_decision` to present
|
|
70
|
+
the native dialog. The answer is provided by the owner to the host; never invent human identity
|
|
71
|
+
or pass an answer as a model argument. A native permission does not approve another gate.
|
|
72
|
+
- Delivery: distinguish what was implemented, validated, published and what is still pending.
|
|
73
|
+
|
|
74
|
+
Phases, checks and gates belong to the core. The phase skills and the checklists of the
|
|
75
|
+
catalog come with this entry. If the common skill is missing, report an incomplete
|
|
76
|
+
installation instead of inventing another methodology.
|
|
77
|
+
|
|
78
|
+
Use the project's `ork` executable. If it is not on the session PATH, the installer recorded
|
|
79
|
+
this path: `{{ork_bin}}`. Do not change global configuration to resolve the PATH.
|
|
80
|
+
|
|
81
|
+
Without MCP and with `.git` protected by the sandbox, report a pending installation of the native
|
|
82
|
+
transport; do not repeat the denied mutation, do not create the thread by other means and do not widen the sandbox.
|
|
83
|
+
|
|
84
|
+
No background monitor is created by this skill. Follow along while the session is
|
|
85
|
+
active and leave a checkpoint for resumption, with the last result and next step proven.
|