- Remover profile [ai] do Ollama para rodar sempre em prod
- Remover profile [bi] do Superset para rodar sempre em prod
- Adicionar MiroFlow service com Dockerfile.miroflow
- Configurar Traefik labels para miroflow.onboardbi.com.br
- Adicionar healthcheck para MiroFlow
- MiroFlow depende de Ollama e acessa via http://ollama:11434
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Erro: ReferenceError: Brain is not defined
Solução: Adicionar Brain aos imports de lucide-react
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Adicionar MiroFlow ao array de tools com ícone Brain
- Importar componente MiroFlowControl
- Adicionar tipo 'miroflow' ao ActiveTool union
- Renderizar MiroFlowControl quando activeTool === 'miroflow'
- MiroFlow agora visível no DevCenter como card científico
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- New RunExecutor processes 'run-executor' subtasks
- Executes agent specs via Ollama (deepseek-r1:14b)
- Saves results as artifacts in Blackboard
- Integrated into Blackboard agents registry
- Updated /api/agent-defs/:id/run endpoint to use run-executor
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
When using Groq (very low latency), tasks complete before the 3s polling
interval fires. Now checkAll() runs immediately on effect mount.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Documents all 4 tabs (Design/Assemble/Deploy/Galeria), API endpoints,
DB schema, and Blackboard integration. Fase 6 complete.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Orchestrate Center now shows 3 stat cards (total/ready/deployed),
timestamps on each agent row, and an Execute button for deployed agents.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Each agent row in the Assemble Line now has an expandable panel showing
live Blackboard subtask progress (status icons, agent roles) when a task
is running. statusColor/statusLabel moved to module scope.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Cards now have an Info button that opens a dialog showing the agent's
spec content, mode, creation date, and quick Execute/Fork actions.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
O build falhava com ENOENT pois attached_assets/ estava no .gitignore
mas o Cockpit.tsx importa o ícone. Adicionado PNG placeholder mínimo
ao repo para o build do Coolify funcionar.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- PatternDetector e SkillEmergence: corrigir alias @/db → caminho relativo correto (../../../db)
- PatternDetector e SkillEmergence: corrigir @/shared/schema → ../../../shared/schema
- server/openclaw: substituir arcadiaSkills (inexistente) por skills do schema
- SkillEmergence: substituir axios (não instalado) por fetch nativo (Node 20)
- server/skills/routes.ts: adicionar export registerSkillRoutes() esperado por routes.ts
O build estava quebrado desde a Fase 4 (OpenClaw). Esses erros impediam
o Coolify de gerar o container com o Design Studio na sidebar.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Deploy agora via `git push gitea Servidor:main` → Coolify monitora branch main no Gitea interno.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Novo item "Design Studio" abaixo de Automation Center (ícone PenTool, cor roxa)
- Renderiza DevCenter.tsx com abas Design/Assemble/Deploy/Galeria
- DESIGN_STUDIO.md documenta estado atual e pendências para próximo agente
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- CLAUDE.md: remover branch inexistente (claude/analyze-project-0mXjP),
documentar branch Servidor, remote origin, comando de deploy real
- DevelopmentModule: adicionar h-full no flex-1 para habilitar scroll
na view home (conteúdo estava cortado sem scrollbar)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Mantém Fase 4 Sprint 1+2 do local (OpenClaw backend + frontend)
- Incorpora do remoto: Fase 3 MiroFlow, Fase 4 melhorias, Fase 5 Automation Fabric
- Resolve conflito openclaw~origin_Servidor (mantido local)
- Histórias unidas com --allow-unrelated-histories -X ours
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Novo item "Automation Center" na sidebar do DevelopmentModule,
ao lado de Skills Emergentes (OpenClaw). Renderiza AutomationCenter
inline como os outros painéis.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Backend:
- server/automation-fabric/runtimes/: WorkflowEngine, RuleEngine, AgentExecutor,
ScheduleEngine, EventEngine — cada um delega para serviços existentes
- server/automation-fabric/service.ts: AutomationFabricService agrega Central + XOS
virtualmente (zero migração de dados)
- server/automation-fabric/routes.ts: GET /api/automation-fabric/list,
POST /:id/run, PATCH /:id/toggle
Frontend:
- AutomationCenter.tsx em /automations-center: lista unificada Central+XOS,
badge de fonte (Central=azul, XOS=roxo), filtros por fonte/status/busca,
toggle enable/disable e execução manual inline
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Tab "Sugestões" na view switcher da página /skills
- Renderiza OpenClawPanel com lista de skills emergentes pendentes
- Badge amarelo mostra contagem de sugestões pendentes (atualiza a cada 5min)
- Query /api/openclaw/suggestions já conectada ao PatternDetector backend
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- miroflow_service.py: extrai resposta de summary, final_boxed_answer,
exceed_max_turn_summary ou último message_history do assistente
- Troca deepseek-r1:14b por llama3.2:3b nos agentes Statistician e Fiscal Auditor
- shared/schema.ts: adiciona tabelas OpenClaw (detectedPatterns, skillSuggestions)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- PatternDetector (cron horário): analisa skill_executions, detecta padrões
(≥3x, 30 dias, ≥80% confiança), grava em detected_patterns
- Geração de DRAFT skills via llama3.1:8b ao atingir threshold
- API routes: GET /api/openclaw/suggestions, POST accept/reject, GET /api/openclaw/patterns
- OpenClawPanel.tsx: painel de visualização com tabs Sugestões/Padrões
- Aba "Skills Emergentes" (ícone Sparkles) na sidebar do /development
- Migration SQL para detected_patterns + skill_suggestions
- fix: corrige import @/lib/db → ../../db/index em versioning.ts
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- useAgentEmergence.ts: hook para Socket.IO + estado de sugestões
- SkillSuggestion.tsx: modal com preview de skill (padrão, nome, descrição, automação)
- OpenClawWidget.tsx: widget flutuante no canto inferior direito
- App.tsx: integração do OpenClawWidget no layout global
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Remove extra_hosts (não funciona pois Ollama só escuta em 127.0.0.1)
- Adiciona rede ollama-net ao serviço miroflow
- OLLAMA_BASE_URL aponta para nome do container Ollama
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Dockerfile.miroflow: imagem Python 3.11 com miroflow instalado via submodule
- docker-compose.prod.yml: novo serviço `miroflow` (porta 8006, rede interna)
- MIROFLOW_HOST atualizado de host fixo (10.0.7.1) para container `miroflow`
- miroflow_service.py: passa system_prompt e initial_user_message no AgentContext
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Toggle for scientific mode with Brain icon
- Agent selector (Statistician, Fiscal Auditor, Researcher)
- Task textarea with dynamic placeholder per agent
- Analysis button with loading state (Loader2 animation)
- Result display with metadata (model, duration, execution_id)
- Error handling with readable messages
- Styled for dark theme matching BiWorkspace
- Adicionar hashes de commit ao SUMMARY.md
- Documentar fallback llama3.2:3b para researcher
- Documentar conftest.py como desvio Rule 2
- Atualizar STATE.md com plano 03-01 concluido
- Atualizar ROADMAP.md progresso fase 3
- FastAPI microservico porta 8006 com agentes statistician, fiscal_auditor, researcher
- Configuracao de modelos: deepseek-r1:14b (stat/fiscal), llama3.1:8b com fallback 3b (researcher)
- OLLAMA_BASE_URL/v1 via UnifiedOpenAIClient
- test_miroflow_service.py: 6 testes cobrindo REQ-3.1 a REQ-3.4 (todos passam)
- sys.path.insert para submodule MiroFlow sem modificar o submodulo
- Adiciona import de registerMiroFlowRoutes
- Chama registerMiroFlowRoutes(app) após registerBiEngineRoutes(app)
- POST /api/miroflow/analyze e GET /api/miroflow/health agora disponíveis
REQ-3.5, REQ-3.7 — Task 2/2 do plano 03-02
- server/miroflow/engine-proxy.ts: proxy para :8006 com MIROFLOW_TIMEOUT=300_000ms
- server/miroflow/routes.ts: re-export de registerMiroFlowRoutes
- POST /api/miroflow/analyze requer autenticação (req.isAuthenticated())
- GET /api/miroflow/health sem autenticação
- registerExecutionInKG com auditHash SHA-256 via createNode
- KG failure não bloqueia resposta ao cliente
REQ-3.5, REQ-3.7 — Task 1/2 do plano 03-02
Servidor:
- GET /api/skills/marketplace: lista skills namespace=system/active,
marca quais já foram importadas pelo tenant (campo imported)
- POST /api/skills/marketplace/:id/import: clona skill system→tenant
com extends appontando para a origem
UI (Skills.tsx):
- Toggle "Minhas Skills" / "Biblioteca" no header
- Grid de cards 2 colunas com nome, slug, descrição, tags clicáveis,
trigger type e versão
- Botão "Importar" → feedback imediato (badge "Importada" + highlight verde)
- Filtro por busca e tag na Biblioteca
- Estado vazio com instrução para admins publicarem skills system
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- App.tsx: adiciona ProtectedRoute path="/skills"
- Skills.tsx: completion provider para referências no body editor
- dispara ao digitar "/" em qualquer ponto do markdown
- sugere todos os prefixos: /skill/, /skill:system/, /var/, /data/,
/kg/, /file/, /tool/, /agent/
- ao digitar /skill/ lista skills existentes do tenant com slug/desc
- usa ref para evitar closure stale com skillsData
- cleanup do provider no unmount
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Jonas Pacheco
engenharia-creator
Arcádia Dev