sinapse-ai 1.19.2 → 1.20.0

package/.sinapse-ai/product/templates/ide-rules/claude-rules.md

@@ -1,6 +1,6 @@
 # SINAPSE Development Rules for Claude Code
 
-You are working with SINAPSE, an AI-Orchestrated System for Full Stack Development.
+SINAPSE orquestra agentes de IA em squads para fluxos de desenvolvimento complexos.
 
 <!-- SINAPSE-MANAGED-START: core-framework -->
 ## Core Framework Understanding
@@ -11,13 +11,9 @@ SINAPSE is a meta-framework that orchestrates AI agents to handle complex develo
 <!-- SINAPSE-MANAGED-START: constitution -->
 ## Constitution
 
-O SINAPSE possui uma **Constitution formal** com princípios inegociáveis e gates automáticos.
+O SINAPSE possui uma **Constitution formal** com principios inegociaveis (`.sinapse-ai/constitution.md`). Gates automaticos bloqueiam violacoes no caminho autonomo (pipeline/motor); no chat interativo a delegacao (Art. VIII) e instrucao de prompt, nao bloqueio de hook — ver Art. VIII na Constitution para a lista exata do que e deterministico.
 
-**Documento completo:** `.sinapse-ai/constitution.md`
-
-**Princípios fundamentais:**
-
-| Artigo | Princípio | Severidade |
+| Artigo | Principio | Severidade |
 |--------|-----------|------------|
 | I | CLI First | NON-NEGOTIABLE |
 | II | Agent Authority | NON-NEGOTIABLE |
@@ -29,334 +25,109 @@ O SINAPSE possui uma **Constitution formal** com princípios inegociáveis e gat
 | VIII | Mandatory Delegation | NON-NEGOTIABLE |
 | IX | Safe Collaboration | NON-NEGOTIABLE |
 | X | Security & Data Protection | NON-NEGOTIABLE |
-
-**Gates automáticos bloqueiam violações.** Consulte a Constitution para detalhes completos.
+| XI | Conservative Default | MUST |
 <!-- SINAPSE-MANAGED-END: constitution -->
 
 <!-- SINAPSE-MANAGED-START: sistema-de-agentes -->
 ## Sistema de Agentes
 
-### Ativação de Agentes
-Use `@agent-name` ou `/SINAPSE:agents:agent-name`:
-
-| Agente | Persona | Escopo Principal |
-|--------|---------|------------------|
-| `@developer` | Dex | Implementação de código |
-| `@quality-gate` | Quinn | Testes e qualidade |
-| `@architect` | Aria | Arquitetura e design técnico |
-| `@project-lead` | Morgan | Product Management |
-| `@product-lead` | Pax | Product Owner, stories/epics |
-| `@sprint-lead` | River | Scrum Master |
-| `@analyst` | Alex | Pesquisa e análise |
-| `@data-engineer` | Dara | Database design |
-| `@ux-design-expert` | Uma | UX/UI design |
-| `@devops` | Gage | CI/CD, git push (EXCLUSIVO) |
-
-### Comandos de Agentes
-Use prefixo `*` para comandos:
-- `*help` - Mostrar comandos disponíveis
-- `*create-story` - Criar story de desenvolvimento
-- `*task {name}` - Executar task específica
-- `*exit` - Sair do modo agente
+Ative com `@agent-name`; comandos usam prefixo `*`.
+
+| Agente | Persona | Escopo |
+|--------|---------|--------|
+| `@developer` | Pixel | Codigo |
+| `@quality-gate` | Litmus | Qualidade |
+| `@architect` | Stratum | Arquitetura |
+| `@project-lead` | Beacon | Product Management |
+| `@product-lead` | Axis | Product Owner |
+| `@sprint-lead` | Sync | Scrum Master |
+| `@analyst` | Scope | Pesquisa |
+| `@data-engineer` | Tensor | Database |
+| `@ux-design-expert` | Mosaic | UX/UI |
+| `@devops` | Pipeline | CI/CD, git push (EXCLUSIVO) |
 <!-- SINAPSE-MANAGED-END: sistema-de-agentes -->
 
 <!-- SINAPSE-MANAGED-START: agent-system -->
 ## Agent System
 
-### Agent Activation
-- Agents are activated with @agent-name syntax: @developer, @quality-gate, @architect, @project-lead, @product-lead, @sprint-lead, @analyst
-- The master agent is activated with @sinapse-orqx
-- Agent commands use the * prefix: *help, *create-story, *task, *exit
-
-### Agent Context
-When an agent is active:
-- Follow that agent's specific persona and expertise
-- Use the agent's designated workflow patterns
-- Maintain the agent's perspective throughout the interaction
+Master agent: `@sinapse-orqx`. Stay in persona until `*exit`.
 <!-- SINAPSE-MANAGED-END: agent-system -->
 
 ## Development Methodology
 
-### Documentation-First Development (NON-NEGOTIABLE)
-1. **Documentation pipeline is AUTOMATIC** - Every briefing triggers: Epic → Story → Validation → Implementation
-2. **No code without validated story** - Story must be in `docs/stories/` with status >= Ready
-3. **Update progress** - Mark checkboxes as tasks complete: [ ] → [x]
-4. **Track changes** - Maintain the File List section in the story
-5. **Follow criteria** - Implement exactly what the acceptance criteria specify
-6. **Delegation is mandatory** - Orchestrators NEVER execute domain work, they ALWAYS delegate to specialists
-
-### Code Standards
-- Write clean, self-documenting code
-- Follow existing patterns in the codebase
-- Include comprehensive error handling
-- Add unit tests for all new functionality
-- Use TypeScript/JavaScript best practices
-
-### Testing Requirements
-- Run all tests before marking tasks complete
-- Ensure linting passes: `npm run lint`
-- Verify type checking: `npm run typecheck`
-- Add tests for new features
-- Test edge cases and error scenarios
+Documentation-First (NON-NEGOTIABLE): Epic -> Story -> Validation -> Implementation, automatico. No code sem story em `docs/stories/` com status >= Ready.
 
 <!-- SINAPSE-MANAGED-START: framework-structure -->
 ## SINAPSE Framework Structure
 
-```
-sinapse-ai/
-├── agents/         # Agent persona definitions (YAML/Markdown)
-├── tasks/          # Executable task workflows
-├── workflows/      # Multi-step workflow definitions
-├── templates/      # Document and code templates
-├── checklists/     # Validation and review checklists
-└── rules/          # Framework rules and patterns
-
-docs/
-├── stories/        # Development stories (numbered)
-├── prd/            # Product requirement documents
-├── architecture/   # System architecture documentation
-└── guides/         # User and developer guides
-```
+Framework core em `.sinapse-ai/`; trabalho do projeto em `docs/` (stories, prd, architecture).
 <!-- SINAPSE-MANAGED-END: framework-structure -->
 
 <!-- SINAPSE-MANAGED-START: framework-boundary -->
 ## Framework vs Project Boundary
 
-O SINAPSE usa um modelo de 4 camadas (L1-L4) para separar artefatos do framework e do projeto. Deny rules em `.claude/settings.json` reforçam isso deterministicamente.
-
-| Camada | Mutabilidade | Paths | Notas |
-|--------|-------------|-------|-------|
-| **L1** Framework Core | NEVER modify | `.sinapse-ai/core/`, `.sinapse-ai/constitution.md`, `bin/sinapse.js`, `bin/sinapse-init.js` | Protegido por deny rules |
-| **L2** Framework Templates | NEVER modify | `.sinapse-ai/development/tasks/`, `.sinapse-ai/development/templates/`, `.sinapse-ai/development/checklists/`, `.sinapse-ai/development/workflows/`, `.sinapse-ai/infrastructure/` | Extend-only |
-| **L3** Project Config | Mutable (exceptions) | `.sinapse-ai/data/`, `agents/*/MEMORY.md`, `core-config.yaml` | Allow rules permitem |
-| **L4** Project Runtime | ALWAYS modify | `docs/stories/`, `packages/`, `squads/`, `tests/` | Trabalho do projeto |
+Modelo de 4 camadas (L1-L4) separa framework e projeto (`.claude/settings.json` reforca via deny rules).
 
-**Toggle:** `core-config.yaml` → `boundary.frameworkProtection: true/false` controla se deny rules são ativas (default: true para projetos, false para contribuidores do framework).
+| Camada | Mutabilidade | Exemplo |
+|--------|-------------|---------|
+| **L1** Framework Core | NEVER modify | `.sinapse-ai/core/` |
+| **L2** Framework Templates | NEVER modify | `.sinapse-ai/development/tasks/` |
+| **L3** Project Config | Mutable (exceptions) | `core-config.yaml` |
+| **L4** Project Runtime | ALWAYS modify | `docs/stories/`, `packages/` |
 
-> **Referência formal:** `.claude/settings.json` (deny/allow rules), `.claude/rules/agent-authority.md`
+**Toggle:** `boundary.frameworkProtection` em `core-config.yaml` (default true para projetos, false para contribuidores do framework).
 <!-- SINAPSE-MANAGED-END: framework-boundary -->
 
 <!-- SINAPSE-MANAGED-START: rules-system -->
 ## Rules System
 
-O SINAPSE carrega regras contextuais de `.claude/rules/` automaticamente. Regras com frontmatter `paths:` só carregam quando arquivos correspondentes são editados.
-
-| Rule File | Description |
-|-----------|-------------|
-| `agent-authority.md` | Agent delegation matrix and exclusive operations |
-| `agent-handoff.md` | Agent switch compaction protocol for context optimization |
-| `agent-memory-imports.md` | Agent memory lifecycle and CLAUDE.md ownership |
-| `coderabbit-integration.md` | Automated code review integration rules |
-| `ids-principles.md` | Incremental Development System principles |
-| `mcp-usage.md` | MCP server usage rules and tool selection priority |
-| `story-lifecycle.md` | Story status transitions and quality gates |
-| `workflow-execution.md` | 4 primary workflows (SDC, QA Loop, Spec Pipeline, Brownfield) |
+Carregadas de `.claude/rules/` automaticamente:
 
-> **Diretório:** `.claude/rules/` — rules são carregadas automaticamente pelo Claude Code quando relevantes.
+`agent-authority.md` · `agent-handoff.md` · `agent-memory-imports.md` · `coderabbit-integration.md` · `ids-principles.md` · `mcp-usage.md` · `story-lifecycle.md` · `workflow-execution.md`
 <!-- SINAPSE-MANAGED-END: rules-system -->
 
 <!-- SINAPSE-MANAGED-START: code-intelligence -->
 ## Code Intelligence
 
-O SINAPSE possui um sistema de code intelligence opcional que enriquece operações com dados de análise de código.
-
-| Status | Descrição | Comportamento |
-|--------|-----------|---------------|
-| **Configured** | Provider ativo e funcional | Enrichment completo disponível |
-| **Fallback** | Provider indisponível | Sistema opera normalmente sem enrichment — graceful degradation |
-| **Disabled** | Nenhum provider configurado | Funcionalidade de code-intel ignorada silenciosamente |
-
-**Graceful Fallback:** Code intelligence é sempre opcional. `isCodeIntelAvailable()` verifica disponibilidade antes de qualquer operação. Se indisponível, o sistema retorna o resultado base sem modificação — nunca falha.
-
-**Diagnóstico:** `sinapse doctor` inclui check de code-intel provider status.
-
-> **Referência:** `.sinapse-ai/core/code-intel/` — provider interface, enricher, client
+Opcional: **Configured** (ativo), **Fallback** (indisponivel, degrada), **Disabled** (sem provider). `isCodeIntelAvailable()` verifica antes de operar — nunca falha se ausente.
 <!-- SINAPSE-MANAGED-END: code-intelligence -->
 
 <!-- SINAPSE-MANAGED-START: graph-dashboard -->
 ## Graph Dashboard
 
-O CLI `sinapse graph` visualiza dependências, estatísticas de entidades e status de providers.
-
-### Comandos
-
 ```bash
-sinapse graph --deps                        # Dependency tree (ASCII)
-sinapse graph --deps --format=json          # Output como JSON
-sinapse graph --deps --format=html          # Interactive HTML (abre browser)
-sinapse graph --deps --format=mermaid       # Mermaid diagram
-sinapse graph --deps --format=dot           # DOT format (Graphviz)
-sinapse graph --deps --watch                # Live mode com auto-refresh
-sinapse graph --deps --watch --interval=10  # Refresh a cada 10 segundos
-sinapse graph --stats                       # Entity stats e cache metrics
+sinapse graph --deps                        # Dependency tree
+sinapse graph --deps --format=json          # JSON
+sinapse graph --deps --format=html          # Interactive HTML
+sinapse graph --deps --watch                # Auto-refresh
+sinapse graph --stats                       # Entity stats
 ```
-
-**Formatos de saída:** ascii (default), json, dot, mermaid, html
-
-> **Referência:** `.sinapse-ai/core/graph-dashboard/` — CLI, renderers, data sources
 <!-- SINAPSE-MANAGED-END: graph-dashboard -->
 
 ## Workflow Execution
 
-### Task Execution Pattern
-1. Read the complete task/workflow definition
-2. Understand all elicitation points
-3. Execute steps sequentially
-4. Handle errors gracefully
-5. Provide clear feedback
-
-### Interactive Workflows
-- Workflows with `elicit: true` require user input
-- Present options clearly
-- Validate user responses
-- Provide helpful defaults
+Leia a task completa, execute os passos em sequencia, trate erros com mensagens acionaveis.
 
 ## Best Practices
 
-### When implementing features:
-- Check existing patterns first
-- Reuse components and utilities
-- Follow naming conventions
-- Keep functions focused and testable
-- Document complex logic
-
-### When working with agents:
-- Respect agent boundaries
-- Use appropriate agent for each task
-- Follow agent communication patterns
-- Maintain agent context
-
-### When handling errors:
-```javascript
-try {
-  // Operation
-} catch (error) {
-  console.error(`Error in ${operation}:`, error);
-  // Provide helpful error message
-  throw new Error(`Failed to ${operation}: ${error.message}`);
-}
-```
-
-## Git & GitHub Integration
-
-### Commit Conventions
-- Use conventional commits: `feat:`, `fix:`, `docs:`, `chore:`, etc.
-- Reference story ID: `feat: implement IDE detection [Story 2.1]`
-- Keep commits atomic and focused
-
-### GitHub CLI Usage
-- Ensure authenticated: `gh auth status`
-- Use for PR creation: `gh pr create`
-- Check org access: `gh api user/memberships`
+Verifique padroes existentes, reutilize componentes, rode `npm run lint` + `npm run typecheck` antes de concluir.
 
 <!-- SINAPSE-MANAGED-START: sinapse-patterns -->
 ## SINAPSE-Specific Patterns
 
-### Working with Templates
-```javascript
-const template = await loadTemplate('template-name');
-const rendered = await renderTemplate(template, context);
-```
-
-### Agent Command Handling
-```javascript
-if (command.startsWith('*')) {
-  const agentCommand = command.substring(1);
-  await executeAgentCommand(agentCommand, args);
-}
-```
-
-### Story Updates
-```javascript
-// Update story progress
-const story = await loadStory(storyId);
-story.updateTask(taskId, { status: 'completed' });
-await story.save();
-```
+Templates via `loadTemplate()`/`renderTemplate()`; comandos `*` roteados por `executeAgentCommand()`.
 <!-- SINAPSE-MANAGED-END: sinapse-patterns -->
 
-## Environment Setup
-
-### Required Tools
-- Node.js 18+
-- GitHub CLI
-- Git
-- Your preferred package manager (npm/yarn/pnpm)
+## Debugging
 
-### Configuration Files
-- `.sinapse/config.yaml` - Framework configuration
-- `.env` - Environment variables
-- `sinapse.config.js` - Project-specific settings
+Erros devem incluir contexto acionavel e sugestao de correcao. Rode `sinapse doctor` para diagnosticar o ambiente.
 
 <!-- SINAPSE-MANAGED-START: common-commands -->
 ## Common Commands
 
-### SINAPSE Master Commands
-- `*help` - Show available commands
-- `*create-story` - Create new story
-- `*task {name}` - Execute specific task
-- `*workflow {name}` - Run workflow
-
-### Development Commands
-- `npm run dev` - Start development
-- `npm test` - Run tests
-- `npm run lint` - Check code style
-- `npm run build` - Build project
+- `*help` / `*create-story` / `*task {name}` / `*workflow {name}`
+- `npm run dev` / `npm test` / `npm run lint` / `npm run build`
 <!-- SINAPSE-MANAGED-END: common-commands -->
 
-## Debugging
-
-### Enable Debug Mode
-```bash
-export SINAPSE_DEBUG=true
-```
-
-### View Agent Logs
-```bash
-tail -f .sinapse/logs/agent.log
-```
-
-### Trace Workflow Execution
-```bash
-npm run trace -- workflow-name
-```
-
-## Claude Code Specific Configuration
-
-### Performance Optimization
-- Prefer batched tool calls when possible for better performance
-- Use parallel execution for independent operations
-- Cache frequently accessed data in memory during sessions
-
-### Tool Usage Guidelines
-- Always use the Grep tool for searching, never `grep` or `rg` in bash
-- Use the Task tool for complex multi-step operations
-- Batch file reads/writes when processing multiple files
-- Prefer editing existing files over creating new ones
-
-### Session Management
-- Track story progress throughout the session
-- Update checkboxes immediately after completing tasks
-- Maintain context of the current story being worked on
-- Save important state before long-running operations
-
-### Error Recovery
-- Always provide recovery suggestions for failures
-- Include error context in messages to user
-- Suggest rollback procedures when appropriate
-- Document any manual fixes required
-
-### Testing Strategy
-- Run tests incrementally during development
-- Always verify lint and typecheck before marking complete
-- Test edge cases for each new feature
-- Document test scenarios in story files
-
-### Documentation
-- Update relevant docs when changing functionality
-- Include code examples in documentation
-- Keep README synchronized with actual behavior
-- Document breaking changes prominently
-
 ---
-*SINAPSE Claude Code Configuration v2.0*
+*SINAPSE Claude Code Configuration v5.0*

package/.sinapse-ai/product/templates/statusline/agent-badges.json

@@ -97,7 +97,7 @@
   "claude-code-mastery/swarm-orqx": {
     "emoji": "🕸️",
     "area": "CLAUDE",
-    "name": "Nexus"
+    "name": "Relay"
   },
   "squad-animations/animation-interpreter": {
     "emoji": "🔮",
@@ -332,7 +332,7 @@
   "squad-content/content-orqx": {
     "emoji": "🔗",
     "area": "CONTENT",
-    "name": "Nexus"
+    "name": "Bulletin"
   },
   "squad-content/editorial-strategist": {
     "emoji": "🧭",

package/.sinapse-ai/product/templates/statusline/statusline-script.js

@@ -39,7 +39,6 @@ function buildStatusLine(data) {
   const white = '\x1b[37m';
   const orange = '\x1b[38;5;208m';
   const boldBlue = `${bold}${blue}`;
-  const boldRed = `${bold}${red}`;
 
   const parts = [];
 
@@ -121,9 +120,12 @@ function buildStatusLine(data) {
     parts.push(`${white}\ud83d\udd22 ${msgCount}${reset}`);
   }
 
-  // 10. Alert >200k
+  // 10. Long-context boundary marker (>200k) \u2014 informational, not an error.
+  // Crossing 200k enters long-context/pricing territory for 1M-window
+  // models; expected behavior on large windows, not a danger signal.
+  // Re-signified by Story onda2-p4 (audit AF-20260702 item 2.12).
   if (data.exceeds_200k_tokens) {
-    parts.push(`${boldRed}\u26a0 >200k!${reset}`);
+    parts.push(`${cyan}\u2139 >200k long-context${reset}`);
   }
 
   return parts.join(' | ');

package/.sinapse-ai/workflow-intelligence/engine/wave-analyzer.js

@@ -4,6 +4,14 @@
  * @story WIS-4 - Wave Analysis Engine
  * @version 1.0.0
  *
+ * @remarks No reachable production caller (require-graph verified 2026-07-02): the only
+ * consumer is the orphaned `wave-executor` (.sinapse-ai/core/execution/wave-executor.js),
+ * which itself has zero callers outside its own test. Fate tied to the DEC-03 batch
+ * decision (reserve marcada) — see
+ * docs/epics/epic-ultra-optimization/decisions/DEC-02-workflow-intelligence.md.
+ * The sibling `suggestion-engine` + `learning/` half of this package is NOT affected —
+ * it has a real consumer via the `*next`/`*patterns` tasks.
+ *
  * @example
  * const { WaveAnalyzer } = require('./wave-analyzer');
  * const analyzer = new WaveAnalyzer();

package/CHANGELOG.md

@@ -1,4 +1,33 @@
-# Changelog
+## [1.20.0](https://github.com/caioimori/sinapse-ai/compare/v1.19.2...1.20.0) (2026-07-03)
+
+### Features
+
+* **cli:** spec/plan de 1ª classe + orchestrate no binário canônico + QA honesto no Windows ([#328](https://github.com/caioimori/sinapse-ai/issues/328)) ([6e55337](https://github.com/caioimori/sinapse-ai/commit/6e553373815c75e766fc4c1f002dcd8ae63b5fb6)), closes [S1/#321](https://github.com/S1/sinapse-ai/issues/321)
+* **installer:** motor de contexto ativo pós-install — .synapse/ no caminho canônico + models no template ([#334](https://github.com/caioimori/sinapse-ai/issues/334)) ([6dbe799](https://github.com/caioimori/sinapse-ai/commit/6dbe799fb0a9724bd53b07f0183d5f627c1cd14a))
+* **synapse:** dieta de contexto — Constitution completa só no 1º prompt + budgets honestos ([#323](https://github.com/caioimori/sinapse-ai/issues/323)) ([39b78e2](https://github.com/caioimori/sinapse-ai/commit/39b78e2c0e30746d433670764ec57e34c0691989))
+
+### Bug Fixes
+
+* **cli:** binário canônico honesto — agents/ideate, uninstall hooksPath, PlanTracker, simulate stub ([#321](https://github.com/caioimori/sinapse-ai/issues/321)) ([19b4dd7](https://github.com/caioimori/sinapse-ai/commit/19b4dd7d6aab9a4abf5b1fde754646491ba04362))
+* **context:** dual-trigger de compactação + statusline re-significado pra era 1M ([#329](https://github.com/caioimori/sinapse-ai/issues/329)) ([18f8582](https://github.com/caioimori/sinapse-ai/commit/18f85821693c90792d7d1ffb74e4023319fdcbed))
+* **models:** registry na era Fable 5 + token-economy sem pin de versão ([#322](https://github.com/caioimori/sinapse-ai/issues/322)) ([b1d120c](https://github.com/caioimori/sinapse-ai/commit/b1d120cacf30f7fa9e45bc556d15a663f7c2e869))
+* **models:** varredura final de era — atlas sincronizado + squad mastery sem pins ([#326](https://github.com/caioimori/sinapse-ai/issues/326)) ([bbf4b9b](https://github.com/caioimori/sinapse-ai/commit/bbf4b9b29aac8bd06eb79cc45f7bbd94a198f4b8))
+
+### Documentation
+
+* **decisions:** executa DEC-02/DEC-05 + triagem de segurança na mesa de decisões ([#335](https://github.com/caioimori/sinapse-ai/issues/335)) ([108e6e2](https://github.com/caioimori/sinapse-ai/commit/108e6e2ff987190edf3dd214c2dd06c8748bf5ed))
+* **decisions:** mesa de decisões da Onda 2 — 5 pareceres de arquitetura + medições armadas ([#333](https://github.com/caioimori/sinapse-ai/issues/333)) ([d2d50c3](https://github.com/caioimori/sinapse-ai/commit/d2d50c3d78d23d5884ea00579d78e502c27bc2d6))
+* **honestidade:** Art. VIII escopado ao caminho real + status executável dos workflows ([#330](https://github.com/caioimori/sinapse-ai/issues/330)) ([c277ff8](https://github.com/caioimori/sinapse-ai/commit/c277ff81438330edf56b68975b081fe776d82419)), closes [agent-exit-hooks.js#onCommandComplete](https://github.com/caioimori/agent-exit-hooks.js/issues/onCommandComplete)
+* vitrine imediata honesta — CHANGELOG, user-guide revisado, limitação 1-story visível, ids por família ([#325](https://github.com/caioimori/sinapse-ai/issues/325)) ([ccd2f8b](https://github.com/caioimori/sinapse-ai/commit/ccd2f8b686fd16cc520838b3cb800ff44195b747)), closes [#318](https://github.com/caioimori/sinapse-ai/issues/318)
+* vitrine profunda honesta — guides no veredito híbrido + README com motor visível + relatório da auditoria ([#327](https://github.com/caioimori/sinapse-ai/issues/327)) ([67889e8](https://github.com/caioimori/sinapse-ai/commit/67889e8ddeed672da91a7612ea0171f9773dcb37))
+
+### Maintenance
+
+* **agents:** piloto de ativação enxuta + 24 colisões de codinome resolvidas (27 renames) ([#332](https://github.com/caioimori/sinapse-ai/issues/332)) ([702a146](https://github.com/caioimori/sinapse-ai/commit/702a146d762f6a54644f6dd7b573d7c777c9dd41))
+* installer bilíngue (troubleshooting i18n) + init sem TTY não crasha ([#331](https://github.com/caioimori/sinapse-ai/issues/331)) ([d4434af](https://github.com/caioimori/sinapse-ai/commit/d4434afded0ee892625f80395989763e5dd393d4))
+* **install:** template CLAUDE.md fiel e enxuto + guard de codinome sem blind spot ([#324](https://github.com/caioimori/sinapse-ai/issues/324)) ([64b5868](https://github.com/caioimori/sinapse-ai/commit/64b586881589cd4dd814fc8c0112029ca7607f2c))
+
+# Changelog
 
 All notable changes to SINAPSE will be documented in this file.
 
@@ -17,7 +46,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Documentation
 
-- **epic** — fecha o épico `orchestration-consolidation` com o **veredito medido do checkpoint "matar ou dobrar"**: HÍBRIDO. Na medição multi-story, o modo nativo entregou 3/3 stories em 64s e 1 chamada, enquanto o motor entregou 1/3 em ~13,5min (a coordenação multi-story corrompeu o estado entre stories). O motor é assumido como **assistente de story isolada** (spec + plano reais); a orquestração autônoma multi-story vira **limitação documentada** (`KNOWN-LIMITATIONS.md`), não promessa. Docs públicos já eram honestos (#318).
+- **epic** — fecha o épico `orchestration-consolidation` com o **veredito medido do checkpoint "matar ou dobrar"**: HÍBRIDO. Na medição multi-story, o modo nativo entregou 3/3 stories em 64s e 1 chamada, enquanto o motor entregou 1/3 em ~13,5min (a coordenação multi-story corrompeu o estado entre stories). O motor é assumido como **assistente de story isolada** (spec + plano reais); a orquestração autônoma multi-story vira **limitação documentada** (`KNOWN-LIMITATIONS.md`), não promessa. Escopo real do #318: atualiza os docs do próprio épico (`docs/epics/`) — a revisão dos guides públicos (user-guide, ade-guide, permission-modes, api-reference) segue pendente e é tratada nas ondas seguintes (Onda 1/Onda 2).
 
 ## [1.19.1] — 2026-06-30 — 🔧 Motor honesto ponta-a-ponta + Windows-safe
 

package/README.md

@@ -423,6 +423,16 @@ npx sinapse-ai chrome-brain install   # Instala browser automation
 
 Todos os comandos são seguros para re-rodar. Para listar agentes ativos depois de instalar, abra o Claude Code ou o Codex CLI e digite `@` para autocompletar.
 
+### Orquestração (avançado)
+
+Além do CLI de instalação acima, o pacote expõe um binário separado (`sinapse`) com um motor de orquestração que gera spec, plano e código para uma story. **Escopo medido: executa 1 story por vez** — orquestração autônoma de múltiplas stories encadeadas foi testada e não é suportada (ver [KNOWN-LIMITATIONS.md](https://github.com/caioimori/sinapse-ai/blob/main/docs/epics/epic-orchestration-consolidation/KNOWN-LIMITATIONS.md)).
+
+```bash
+npx -p sinapse-ai sinapse route "<briefing>"       # Classifica o briefing -> workflow + o que falta
+npx -p sinapse-ai sinapse build "<briefing>"       # Roda o route de forma guiada (--dry-run, --type)
+npx -p sinapse-ai sinapse orchestrate <story-id>   # Gera spec + plano + build para UMA story (--status/--stop/--resume)
+```
+
 ---
 
 ## Contribuindo

package/bin/cli.js

@@ -29,6 +29,11 @@ const KNOWN_COMMANDS = [
   'ideate',
   'agents',
   'chrome-brain',
+  // Story onda2-p3 (audit AF-20260702 item 2.4) — the orchestration engine,
+  // reachable from the canonical binary users actually install.
+  'orchestrate',
+  'spec',
+  'plan',
   'help',
 ];
 const {
@@ -160,6 +165,30 @@ function runRouter() {
       cmdDoctor(doctorOpts).catch(e => { logger.error(`${RED}Erro no doctor:${NC} ${e.message}`); process.exit(1); });
       break;
     }
+    case 'orchestrate':
+    case 'spec':
+    case 'plan': {
+      // Story onda2-p3 (audit AF-20260702 item 2.4) — the orchestration engine
+      // (the one capability validated by the HYBRID verdict) was unreachable
+      // from the canonical binary: users who install as the README teaches
+      // (`npx sinapse-ai ...`) never discover bin/sinapse.js. Delegates via
+      // spawnSync to bin/sinapse.js, which owns the arg parsing — same
+      // single-source-of-truth pattern as `init` (cli→sinapse) and the
+      // S1/#321 `agents`/`ideate` delegation (sinapse→cli).
+      const { spawnSync } = require('child_process');
+      const result = spawnSync(
+        process.execPath,
+        [path.join(__dirname, 'sinapse.js'), ...args],
+        { stdio: 'inherit' },
+      );
+      if (result.error) {
+        logger.error(`${RED}Erro no ${command}:${NC} ${result.error.message}`);
+        logger.error(`Tente: ${CYAN}npx sinapse-ai ${command} --help${NC}`);
+        process.exit(1);
+      }
+      process.exit(result.status ?? 0);
+    }
+    // eslint-disable-next-line no-fallthrough -- process.exit above terminates; same pattern as `init`.
     case 'chrome-brain': {
     // Story 10.13 — chrome-brain is the canonical sub-capability for browser
     // automation. Delegating to the shared chrome-brain-installer module keeps

package/bin/commands/help.js

@@ -19,6 +19,11 @@ function cmdHelp() {
   logger.always('');
   logger.always(`  ${DIM}Funciona em CI / ambientes não-interativos (usa defaults sensatos).${NC}`);
   logger.always('');
+  logger.always(`${BOLD}Motor de orquestração (avançado — 1 story por vez):${NC}\n`);
+  logger.always(`  ${CYAN}npx sinapse-ai spec <story-id>${NC}        Gera a spec REAL de uma story (para antes de plano/build/QA)`);
+  logger.always(`  ${CYAN}npx sinapse-ai plan <story-id>${NC}        Spec + plano de implementação REAL (para antes de build/QA)`);
+  logger.always(`  ${CYAN}npx sinapse-ai orchestrate <story-id>${NC} Pipeline completo de 1 story (${DIM}--dry-run pra prever${NC})`);
+  logger.always('');
   logger.always(`${BOLD}Diagnóstico:${NC}\n`);
   logger.always(`  ${CYAN}npx sinapse-ai status${NC}           Mostra estado da instalação`);
   logger.always(`  ${CYAN}npx sinapse-ai doctor${NC}           Roda health checks (--fix --dry-run --json --deep)`);

package/bin/commands/uninstall.js

@@ -183,6 +183,11 @@ async function cmdUninstall(opts = {}) {
     [path.join(HOME, '.claude', 'commands', 'SINAPSE'), '~/.claude/commands/SINAPSE/'],
     [path.join(BIN_DIR, 'sinapse'), '~/bin/sinapse'],
     [path.join(BIN_DIR, 'sinapse.cmd'), '~/bin/sinapse.cmd'],
+    // Story onda2-p9 — project-level context-engine runtime at cwd (parity
+    // with runUninstall in bin/sinapse.js). Fully SINAPSE-generated
+    // (constitution derived from .sinapse-ai/constitution.md + sessions/
+    // metrics runtime state); regenerated by any future install.
+    [path.join(process.cwd(), '.synapse'), './.synapse/ (context engine runtime)'],
   ];
 
   for (const [p, label] of ownedItems) {

package/bin/postinstall.js

@@ -361,6 +361,13 @@ function stepCreateRuntimeDirs() {
  * UserPromptSubmit context engine actually injects rules. Without this, the
  * engine is inert (the hook silently emits no context). Tolerant + non-critical:
  * the wrapper always exits 0 and the engine degrades gracefully if absent.
+ *
+ * NOTE (Story onda2-p9 / DEC-01 option A): user-facing installs now get
+ * .synapse/ from the CANONICAL path — packages/installer/src/installer/
+ * synapse-runtime-installer.js, wired into the wizard (covers `install` and
+ * `init`). This step remains for the repo-source `npm run setup` flow (it
+ * writes at PROJECT_ROOT, i.e. the package/repo itself — see DEC-01 evidence).
+ * Do NOT delete without a deliberate DEC-03-like decision.
  */
 function stepGenerateSynapse() {
   if (isGlobalInstall()) {