aiox-core 5.0.2 → 5.0.4

package/pro/squads/squad-creator-pro/tasks/create-squad.md

@@ -0,0 +1,933 @@
+# Task: Create Complete Squad
+
+**Task ID:** create-squad
+**Version:** 2.0
+**Execution Type:** Hybrid
+**Purpose:** Create a complete AIOS-FULLSTACK squad through guided elicitation, research, and validation
+**Orchestrator:** @squad-chief
+**Mode:** Incremental (human checkpoints) or YOLO (autonomous)
+**Quality Standard:** AIOS Level (all components meet minimum standards)
+
+**Frameworks Used:**
+- `data/tier-system-framework.md` → Agent tier classification (Phase 3)
+- `data/quality-dimensions-framework.md` → Squad validation (Phase 6)
+- `data/decision-heuristics-framework.md` → Checkpoint logic (Phase 4, 6)
+- `data/executor-matrix-framework.md` → Task executor assignment (Phase 4)
+
+---
+
+## Overview
+
+This task creates a complete, production-ready squad with all required components: agents, tasks, workflows, templates, checklists, and knowledge bases. The key insight: **squads must be built on researched elite minds, not generic bots**.
+
+**Context-first entrypoint (recommended):**
+- Run `detect-squad-context` first.
+- If context is `greenfield_pure`, `pre_existing_brief`, or `partial_squad`, route to `wf-context-aware-create-squad`.
+- If context is `legacy_assets`, route to `wf-brownfield-upgrade-squad`.
+
+**v2.0 Changes:**
+- Mandatory research loop before any agent creation
+- PHASE-based structure with checkpoints
+- Quality gates with blocking requirements
+- Framework integration at every decision point
+
+**Token economy rule (commands):**
+- Keep primary commands only in `commands`.
+- Store PT-BR aliases once in `command_aliases_ptbr`.
+- Do not duplicate aliases in `commands`, `command_visibility`, and `all_commands`.
+
+```
+INPUT (domain + purpose + target_user)
+    ↓
+[PHASE 0: DISCOVERY]
+    → Validate domain viability
+    → Check for existing similar squads
+    ↓
+[PHASE 1: RESEARCH]
+    → Execute mind-research-loop.md
+    → Identify elite minds with frameworks
+    → Validate each mind (SC_FV_001)
+    ↓
+[PHASE 2: ARCHITECTURE]
+    → Define tier structure
+    → Plan agent relationships
+    → Design quality gates
+    ↓
+[PHASE 3: CREATION]
+    → Create agents (wf-research-then-create-agent.yaml)
+    → Create workflows with checkpoints
+    → Create tasks with Task Anatomy
+    ↓
+[PHASE 4: INTEGRATION]
+    → Create orchestrator agent
+    → Wire dependencies
+    → Generate documentation
+    ↓
+[PHASE 5: VALIDATION]
+    → Run squad-checklist.md
+    → Quality dimensions scoring
+    → Fix blocking issues
+    ↓
+OUTPUT: Complete squad + Quality Score
+```
+
+---
+
+## Inputs
+
+| Parameter | Type | Required | Description | Example |
+|-----------|------|----------|-------------|---------|
+| `domain` | string | Yes | Domain expertise area | `"copywriting"`, `"legal"`, `"data"` |
+| `purpose` | string | Yes | What the squad should accomplish | `"Create high-converting sales pages"` |
+| `target_user` | string | Yes | Who will use this squad | `"Marketing teams at SaaS companies"` |
+| `use_cases` | list | Yes | 3-5 key use cases | `["sales pages", "email sequences", "ads"]` |
+| `mode` | enum | Yes | `"incremental"` or `"yolo"` | `"incremental"` |
+| `squad_name` | string | No | Override default name | `"copy"` |
+
+---
+
+## Preconditions
+
+- [ ] squad-chief agent is active
+- [ ] WebSearch tool available (for research)
+- [ ] Write permissions for `squads/` directory
+- [ ] Frameworks loaded: tier-system, quality-dimensions, decision-heuristics, executor-matrix
+
+---
+
+## PHASE 0: DISCOVERY
+
+**Duration:** 5-10 minutes
+**Checkpoint:** SC_DSC_001 (Discovery Complete)
+**Mode:** Interactive (both modes)
+
+### Step 0.0: Workspace Domain Awareness
+
+**Actions:**
+```yaml
+workspace_domain_check:
+  read_files:
+    - "workspace/workspace.yaml"          # Lista de domínios e providers
+    - "workspace/domains/"                # Domínios existentes
+    - "workspace/products/"               # Produtos existentes
+
+  checks:
+    - domain_exists_in_workspace: "workspace/domains/{domain}/ existe?"
+    - product_alignment: "Squad serve qual produto em workspace/products/?"
+    - provider_overlap: "Squad usa providers já declarados em workspace.yaml?"
+
+  output:
+    workspace_context:
+      domain_match: "exact | partial | none"
+      related_products: ["list"]
+      available_providers: ["list"]
+      existing_squads_for_domain: ["list"]
+
+  behavior:
+    domain_exists: "Informar ao usuário que o domínio já tem definições no workspace. Squad deve alinhar com entities.yaml e workflows.yaml do domínio."
+    domain_not_exists: "Prosseguir normalmente. Após criação, sugerir criação de workspace/domains/{domain}/"
+```
+
+### Step 0.1: Validate Domain Viability
+
+**Actions:**
+```yaml
+domain_viability_check:
+  questions:
+    - "Is this domain well-documented with established methodologies?"
+    - "Do elite practitioners exist with documented frameworks?"
+    - "Can outputs be validated against objective criteria?"
+    - "Is there enough depth to warrant a squad (vs single agent)?"
+
+  scoring:
+    - documented_methodologies: 0-3
+    - elite_practitioners: 0-3
+    - validation_criteria: 0-2
+    - complexity_warrants_squad: 0-2
+
+  threshold: 6/10
+  veto_conditions:
+    - elite_practitioners < 2 → "Cannot create quality agents without researched minds"
+```
+
+**Decision Point:**
+```
+IF viability_score >= 6:
+    → PROCEED to Step 0.2
+ELSE IF viability_score >= 4:
+    → WARN: "Domain may not support full squad. Consider single agent."
+    → ASK: "Proceed with limited scope?"
+ELSE:
+    → BLOCK: "Domain lacks sufficient documented frameworks"
+    → SUGGEST: "Consider researching domain first or choosing adjacent domain"
+```
+
+### Step 0.1.5: Workspace Domain Awareness
+
+**Actions:**
+```yaml
+workspace_domain_check:
+  read_files:
+    - "workspace/workspace.yaml"          # Domínios e providers registrados
+    - "workspace/domains/"                # Domínios existentes
+    - "workspace/products/"               # Produtos existentes
+
+  analysis:
+    - domain_exists: "workspace/domains/{domain}/ existe?"
+    - domain_has_entities: "workspace/domains/{domain}/entities.yaml existe?"
+    - domain_has_workflows: "workspace/domains/{domain}/workflows.yaml existe?"
+    - related_products: "Quais produtos referenciam este domínio?"
+    - provider_alignment: "Quais providers o domínio usa?"
+
+  output:
+    workspace_context:
+      domain_exists: true | false
+      entities_count: N
+      workflows_count: N
+      related_products: ["product_a", "product_b"]
+      providers: ["clickup", "supabase"]
+    recommendation: "Alinhar agentes do squad com entidades e workflows existentes no workspace"
+```
+
+**Usage:**
+- Se `workspace/domains/{domain}/` existe → informar squad-chief para que agentes respeitem entidades existentes
+- Se `workspace/domains/{domain}/entities.yaml` existe → incluir entidades como contexto nos agentes criados
+- Se `workspace/products/` tem produtos relevantes → alinhar use cases com produtos existentes
+- Informação é **read-only** — não modifica workspace, apenas enriquece o contexto da criação
+
+### Step 0.2: Check Existing Squads
+
+**Actions:**
+```yaml
+existing_squad_check:
+  search_paths:
+    - "squads/{similar_names}/"
+    - "squads/*/{domain}*"
+    - ".claude/commands/*/{domain}*"
+
+  analysis:
+    - existing_coverage: "What does existing squad cover?"
+    - gap_identification: "What's missing?"
+    - extension_vs_new: "Extend existing or create new?"
+
+  output:
+    decision: "extend" | "create_new" | "abort_duplicate"
+    rationale: "..."
+```
+
+**Decision Point:**
+```
+IF existing squad covers 80%+ of use cases:
+    → SUGGEST: "Extend existing squad instead"
+    → IF user confirms: → GOTO extend-squad workflow
+IF partial overlap:
+    → WARN: "Partial overlap detected. Plan for integration."
+ELSE:
+    → PROCEED to Phase 1
+```
+
+### Step 0.3: Define Squad Structure
+
+**Actions:**
+```yaml
+pack_structure_elicitation:
+  required:
+    - squad_name: "kebab-case identifier"
+    - pack_title: "Human-readable title"
+    - version: "1.0.0"
+    - author: "Organization or person"
+    - entry_agent: "Primary activation agent id (default: {squad_name}-chief)"
+    - slash_prefix: "camelCase for commands"
+
+  derived:
+    - pack_path: "squads/{squad_name}/"
+    - outputs_path: "squads/{squad_name}/outputs/"
+    - command_path: ".claude/commands/{PackTitle}/"
+
+  pattern_library:
+    - prefix: "2-letter code (e.g., CP for Copy)"
+    - initial_patterns:
+      - "{PREFIX}-001: Core Process"
+      - "{PREFIX}-002: Quality Standard"
+      - "{PREFIX}-003: Exception Handling"
+```
+
+**Output (PHASE 0):**
+```yaml
+phase_0_output:
+  viability_score: 8/10
+  decision: "create_new"
+  squad_name: "copy"
+  pack_title: "Copy Squad"
+  entry_agent: "copy-chief"
+  slash_prefix: "copy"
+  pattern_prefix: "CP"
+  mode: "incremental"
+```
+
+**Checkpoint SC_DSC_001:**
+```yaml
+heuristic_id: SC_DSC_001
+name: "Discovery Complete"
+blocking: true
+criteria:
+  - viability_score >= 6
+  - squad_name defined
+  - no duplicate squad
+```
+
+---
+
+## PHASE 1: RESEARCH
+
+**Duration:** 15-30 minutes
+**Checkpoint:** SC_RES_001 (Research Quality Gate)
+**Mode:** Autonomous (both modes)
+
+### Step 1.1: Execute Mind Research Loop
+
+**CRITICAL:** This step is MANDATORY. No agent creation without research.
+
+**Actions:**
+```yaml
+execute_workflow:
+  workflow: "workflows/mind-research-loop.md"
+  inputs:
+    domain: "{domain}"
+    min_minds: 5
+    max_minds: 10
+    iterations: 3-5
+
+  execution_mode: "autonomous"
+  human_review: false  # Research loop handles validation internally
+```
+
+**Mind Research Loop Summary:**
+```
+Iteration 1: Broad research (find 15-20 names)
+Iteration 2: Devil's advocate (cut weak candidates)
+Iteration 3: Framework validation (SC_FV_001)
+Output: Curated list of elite minds with documented frameworks
+```
+
+### Step 1.2: Validate Research Output
+
+**Actions:**
+```yaml
+research_validation:
+  # From data/quality-dimensions-framework.md
+  dimensions_to_check:
+    - accuracy: "Are the minds real with verifiable work?"
+    - coherence: "Do the frameworks complement each other?"
+    - strategic_alignment: "Do they cover the use cases?"
+    - operational_excellence: "Is the research actionable?"
+
+  # From data/decision-heuristics-framework.md
+  apply_heuristic:
+    id: SC_RES_001
+    name: "Research Quality Gate"
+    threshold: 7.0
+    blocking: true
+
+    criteria:
+      - minds_count >= 5
+      - all_minds_have_frameworks: true
+      - tier_0_covered: true
+      - use_cases_mapped >= 80%
+
+    veto_conditions:
+      - minds_count < 3 → "Insufficient research"
+      - no_tier_0_minds → "No diagnostic/foundation agent"
+```
+
+**Output (PHASE 1):**
+```yaml
+phase_1_output:
+  minds:
+    - name: "Gary Halbert"
+      tier: 1
+      frameworks: ["A-Pile/B-Pile", "Swipe File Method"]
+      covers_use_cases: ["sales pages", "direct mail"]
+
+    - name: "Eugene Schwartz"
+      tier: 0
+      frameworks: ["Market Awareness", "5 Levels of Sophistication"]
+      covers_use_cases: ["all - diagnostic"]
+
+    # ... more minds
+
+  research_score: 8.5/10
+  checkpoint_status: "PASS"
+```
+
+---
+
+## PHASE 2: ARCHITECTURE
+
+**Duration:** 10-20 minutes
+**Checkpoint:** SC_ARC_001 (Architecture Approved)
+**Mode:** Interactive (incremental) / Autonomous (YOLO)
+
+### Step 2.1: Define Tier Structure
+
+**Apply: tier-system-framework.md**
+
+**Actions:**
+```yaml
+tier_structure_design:
+  # Assign each mind to a tier
+  orchestrator:
+    purpose: "Coordinates all tiers, routes requests"
+    candidates: "New agent or promote from Tier 0"
+
+  tier_0_diagnosis:
+    purpose: "First contact, analysis, classification"
+    minds: "Those who created diagnostic frameworks"
+    required: true  # Every squad MUST have Tier 0
+
+  tier_1_masters:
+    purpose: "Primary experts with proven results"
+    minds: "Those with documented $X results"
+
+  tier_2_systematizers:
+    purpose: "Created frameworks others use"
+    minds: "Thought leaders, framework creators"
+
+  tier_3_specialists:
+    purpose: "Specific format/channel experts"
+    minds: "Niche specialists"
+
+  tools:
+    purpose: "Validation, checklists, calculators"
+    examples: ["quality-checker", "compliance-validator"]
+```
+
+### Step 2.2: Plan Agent Relationships
+
+**Actions:**
+```yaml
+agent_relationships:
+  handoff_map:
+    # Who hands off to whom
+    - from: "orchestrator"
+      to: "tier_0_agents"
+      when: "New request arrives"
+
+    - from: "tier_0_agents"
+      to: "tier_1_agents"
+      when: "Diagnosis complete, execution needed"
+
+  synergies:
+    # Who works well together
+    - agents: ["diagnosis-agent", "master-agent"]
+      pattern: "Diagnosis feeds master context"
+
+  conflicts:
+    # Who should NOT be combined
+    - agents: ["aggressive-style", "conservative-style"]
+      reason: "Contradictory approaches"
+```
+
+### Step 2.3: Design Quality Gates
+
+**Actions:**
+```yaml
+quality_gates_design:
+  gates:
+    - id: "QG-001"
+      name: "Request Classification"
+      transition: "Input → Tier 0"
+      type: "routing"
+      criteria: "Request type identified"
+
+    - id: "QG-002"
+      name: "Diagnosis Complete"
+      transition: "Tier 0 → Tier 1"
+      type: "blocking"
+      criteria: "Analysis approved, requirements clear"
+
+    - id: "QG-003"
+      name: "Draft Review"
+      transition: "Execution → Output"
+      type: "blocking"
+      criteria: "Quality checklist passed"
+
+  escalation_paths:
+    - on_failure: "Return to previous tier with feedback"
+    - on_repeated_failure: "Escalate to human review"
+```
+
+**Output (PHASE 2):**
+```yaml
+phase_2_output:  # [Example]
+  tier_structure:
+    orchestrator: "{squad}-chief"
+    tier_0: ["{diagnosis-agent-1}", "{diagnosis-agent-2}"]
+    tier_1: ["{master-agent-1}", "{master-agent-2}"]
+    tier_2: ["{systematizer-1}", "{systematizer-2}"]
+    tier_3: ["{specialist-1}", "{specialist-2}"]
+    tools: ["{tool-1}", "{tool-2}"]
+
+  quality_gates: 5
+  handoffs: 12
+  architecture_score: 8/10
+```
+
+**Checkpoint SC_ARC_001:**
+```yaml
+heuristic_id: SC_ARC_001
+name: "Architecture Approved"
+blocking: true
+criteria:
+  - tier_0_defined: true
+  - orchestrator_defined: true
+  - quality_gates >= 3
+  - handoff_map_complete: true
+
+# Incremental mode: Human approval required
+# YOLO mode: Auto-approve if criteria met
+```
+
+---
+
+## PHASE 3: CREATION
+
+**Duration:** 30-60 minutes (varies by squad size)
+**Checkpoint:** SC_CRT_001 (Creation Quality)
+**Mode:** Autonomous (both modes, with quality gates)
+
+### Step 3.1: Create Directory Structure
+
+**Actions:**
+```yaml
+create_directories:
+  base: "squads/{squad_name}/"
+  subdirectories:
+    - agents/
+    - tasks/
+    - workflows/
+    - templates/
+    - checklists/
+    - data/
+    - docs/
+    - outputs/
+
+  initial_files:
+    - config.yaml (from templates/config-tmpl.yaml)
+    - README.md (placeholder)
+    - outputs/README.md (runtime artifact location guide)
+```
+
+### Step 3.2: Create Agents (For Each Mind)
+
+**Execute workflow for each mind:**
+```yaml
+for_each_mind:
+  workflow: "workflows/wf-research-then-create-agent.yaml"
+  inputs:
+    agent_purpose: "{derived from mind's expertise}"
+    domain: "{domain}"
+    specialist_slug: "{mind_slug}"
+    specialist_name: "{mind_name}"
+    squad_name: "{squad_name}"
+
+  quality_gate:
+    # Each agent must pass SC_AGT_001
+    blocking: true
+    retry_on_fail: true
+    max_retries: 2
+```
+
+### Step 3.3: Create Orchestrator Agent
+
+**Special agent creation:**
+```yaml
+create_orchestrator:
+  agent_id: "{squad_name}-chief"
+  role: "Squad Orchestrator"
+  tier: "orchestrator"
+
+  special_capabilities:
+    - "Route requests to appropriate tier"
+    - "Manage multi-agent workflows"
+    - "Track context across handoffs"
+    - "Report squad status"
+
+  commands:
+    - "*help" - List all squad capabilities
+    - "*route" - Route request to best agent
+    - "*status" - Show current context
+    - "*handoff" - Transfer to specific agent
+
+  must_include:
+    - tier_routing_logic
+    - quality_gate_checks
+    - context_preservation
+```
+
+### Step 3.4: Create Workflows
+
+**For complex multi-step operations:**
+```yaml
+create_workflows:
+  # Identify operations that need workflows
+  criteria:
+    - spans_multiple_agents: true
+    - has_intermediate_checkpoints: true
+    - output_feeds_next_step: true
+
+  template: "templates/workflow-tmpl.yaml"
+
+  minimum_structure:
+    - phases: 3+
+    - checkpoints_per_phase: 1+
+    - framework_references: 2+
+
+  quality_standard:
+    min_lines: 500
+```
+
+### Step 3.5: Create Tasks
+
+**Apply: executor-matrix-framework.md**
+
+**For atomic single-session operations:**
+```yaml
+create_tasks:
+  # Each task follows Task Anatomy (8 fields)
+  task_anatomy:
+    - id: "Unique identifier"
+    - purpose: "What and why"
+    - executor: "Human | Agent | Hybrid | Worker"
+    - inputs: "What it needs"
+    - preconditions: "What must be true"
+    - steps: "How to execute"
+    - outputs: "What it produces"
+    - validation: "How to verify"
+
+  executor_assignment:
+    # From data/executor-matrix-framework.md
+    decision_tree:
+      - IF rule_based AND low_stakes → Worker
+      - ELSE IF needs_judgment → Hybrid (AI draft, human approve)
+      - ELSE IF creative_synthesis → Agent
+      - ELSE → Human
+
+  quality_standard:
+    simple_tasks: 300+ lines
+    complex_tasks: 500+ lines
+```
+
+**Output (PHASE 3):**
+```yaml
+phase_3_output:
+  agents_created: 8
+  all_pass_quality_gate: true
+  workflows_created: 3
+  tasks_created: 12
+  templates_created: 5
+  total_lines: 14,500
+```
+
+---
+
+## PHASE 4: INTEGRATION
+
+**Duration:** 10-15 minutes
+**Checkpoint:** SC_INT_001 (Integration Complete)
+**Mode:** Autonomous
+
+### Step 4.1: Wire Dependencies
+
+**Actions:**
+```yaml
+wire_dependencies:
+  for_each_agent:
+    - verify tasks exist
+    - verify templates exist
+    - verify checklists exist
+    - update dependencies block
+
+  create_missing:
+    - if dependency referenced but doesn't exist → create stub
+    - mark stubs with TODO for later completion
+```
+
+### Step 4.2: Create Knowledge Base
+
+**Actions:**
+```yaml
+create_knowledge_base:
+  file: "data/{squad_name}-kb.md"
+
+  sections:
+    - domain_overview: "What this domain is about"
+    - key_concepts: "Essential terminology"
+    - best_practices: "From researched minds"
+    - common_patterns: "Reusable patterns"
+    - anti_patterns: "What to avoid"
+    - regulatory: "If applicable"
+```
+
+### Step 4.3: Generate Documentation
+
+**Actions:**
+```yaml
+generate_documentation:
+  readme:
+    file: "README.md"
+    sections:
+      - overview
+      - installation
+      - quick_start
+      - agents_list
+      - workflows_list
+      - tasks_list
+      - templates_list
+      - usage_examples
+
+  config:
+    file: "config.yaml"
+    validate: true
+```
+
+---
+
+## PHASE 5: VALIDATION
+
+**Duration:** 5-10 minutes
+**Checkpoint:** SC_VAL_001 (Squad Validated)
+**Mode:** Autonomous with human review option
+
+### Step 5.1: Run Squad Checklist
+
+**Execute: checklists/squad-checklist.md**
+
+**Actions:**
+```yaml
+run_squad_checklist:
+  categories:
+    - structure: "Directory structure correct"
+    - agents: "All agents meet standards"
+    - workflows: "All workflows have checkpoints"
+    - tasks: "All tasks follow anatomy"
+    - documentation: "README complete"
+    - integration: "Dependencies wired"
+
+  blocking_items:
+    - orchestrator_exists
+    - tier_0_exists
+    - quality_gates_defined
+    - readme_complete
+```
+
+### Step 5.2: Quality Dimensions Scoring
+
+**Apply: quality-dimensions-framework.md**
+
+**Actions:**
+```yaml
+quality_scoring:
+  dimensions:
+    - accuracy: "Are agents based on real minds?"
+    - coherence: "Do components work together?"
+    - strategic_alignment: "Does squad serve purpose?"
+    - operational_excellence: "Is squad usable?"
+    - innovation_capacity: "Can squad grow?"
+    - risk_management: "Are guardrails in place?"
+
+  weights:
+    accuracy: 0.20
+    coherence: 0.20
+    strategic_alignment: 0.15
+    operational_excellence: 0.20
+    innovation_capacity: 0.10
+    risk_management: 0.15
+
+  threshold: 7.0
+  veto_if_below_5:
+    - accuracy
+    - coherence
+    - operational_excellence
+```
+
+### Step 5.3: Fix Blocking Issues
+
+**Actions:**
+```yaml
+fix_blocking_issues:
+  for_each_blocking_issue:
+    - identify_root_cause
+    - determine_fix_approach
+    - execute_fix
+    - re-validate
+
+  max_iterations: 3
+  on_max_iterations_exceeded:
+    - report_remaining_issues
+    - ask_human_for_direction
+```
+
+**Output (PHASE 5):**
+```yaml
+phase_5_output:
+  checklist_pass: true
+  quality_score: 8.2/10
+  blocking_issues_resolved: 2
+  warnings: 3
+  status: "PASS"
+```
+
+---
+
+## PHASE 6: HANDOFF
+
+**Duration:** 2-5 minutes
+**Mode:** Interactive
+
+### Step 6.1: Present Squad Summary
+
+**Actions:**
+```yaml
+present_summary:
+  created:
+    - agents: 8
+    - workflows: 3
+    - tasks: 12
+    - templates: 5
+    - checklists: 4
+
+  quality:
+    - overall_score: 8.2/10
+    - research_foundation: "5 elite minds"
+    - tier_coverage: "Full (0-3 + tools)"
+
+  activation:
+    - install: "npm run install:squad {squad_name}"
+    - activate: "@{squad_name}"
+    - example: "@{squad-name}:{agent-name}"  # e.g., @copy:sales-page-writer
+```
+
+### Step 6.2: Document Next Steps
+
+**Actions:**
+```yaml
+next_steps:
+  optional_improvements:
+    - "Add more specialists to Tier 3"
+    - "Create domain-specific templates"
+    - "Add integration tests"
+
+  handoff_to:
+    - agent: "qa-architect"
+      when: "Need deep validation audit"
+    - agent: "domain-orchestrator"
+      when: "Ready to use squad"
+```
+
+---
+
+## Outputs
+
+| Output | Location | Description |
+|--------|----------|-------------|
+| Squad Directory | `squads/{squad_name}/` | Complete squad structure |
+| Agents | `squads/{squad_name}/agents/` | All agent definitions |
+| Workflows | `squads/{squad_name}/workflows/` | Multi-phase workflows |
+| Tasks | `squads/{squad_name}/tasks/` | Atomic tasks |
+| Templates | `squads/{squad_name}/templates/` | Output templates |
+| Checklists | `squads/{squad_name}/checklists/` | Validation checklists |
+| Runtime Outputs | `squads/{squad_name}/outputs/` | Artifacts gerados em execuções reais |
+| Knowledge Base | `squads/{squad_name}/data/{squad_name}-kb.md` | Domain knowledge |
+| Documentation | `squads/{squad_name}/README.md` | Usage documentation |
+| Config | `squads/{squad_name}/config.yaml` | Squad configuration |
+
+---
+
+## Validation Criteria (All Must Pass)
+
+### Structure
+- [ ] Squad directory exists at `squads/{squad_name}/`
+- [ ] All required subdirectories created
+- [ ] config.yaml is valid YAML
+- [ ] config.yaml defines explicit `entry_agent`
+- [ ] `outputs/` exists with bootstrap artifact (`outputs/README.md` or equivalent)
+
+### Agents
+- [ ] Orchestrator agent exists
+- [ ] At least one Tier 0 agent exists
+- [ ] All agents pass SC_AGT_001 (300+ lines, voice_dna, etc.)
+- [ ] All agents have tier assigned
+
+### Workflows
+- [ ] Complex operations have workflows (not just tasks)
+- [ ] All workflows have 3+ phases
+- [ ] All workflows have checkpoints
+
+### Tasks
+- [ ] All tasks follow Task Anatomy (8 fields)
+- [ ] Complex tasks are 500+ lines
+- [ ] Executor assigned for each task
+
+### Documentation
+- [ ] README.md complete with all sections
+- [ ] Usage examples provided
+- [ ] Installation instructions clear
+
+### Quality
+- [ ] Overall score >= 7.0
+- [ ] No blocking items failed
+- [ ] Research foundation documented
+
+---
+
+## Heuristics Reference
+
+| Heuristic ID | Name | Where Applied | Blocking |
+|--------------|------|---------------|----------|
+| SC_DSC_001 | Discovery Complete | Phase 0 | Yes |
+| SC_RES_001 | Research Quality Gate | Phase 1 | Yes |
+| SC_ARC_001 | Architecture Approved | Phase 2 | Yes |
+| SC_AGT_001 | Agent Quality Gate | Phase 3 | Yes |
+| SC_CRT_001 | Creation Quality | Phase 3 | Yes |
+| SC_INT_001 | Integration Complete | Phase 4 | No |
+| SC_VAL_001 | Squad Validated | Phase 5 | Yes |
+
+---
+
+## Error Handling
+
+```yaml
+error_handling:
+  research_fails:
+    - retry_with_different_queries
+    - if_still_fails: "Abort with explanation"
+
+  agent_creation_fails:
+    - identify_missing_research
+    - supplement_research
+    - retry_creation
+    - if_still_fails: "Create simpler agent, note for improvement"
+
+  validation_fails:
+    - log_specific_failures
+    - attempt_automated_fix
+    - if_cannot_fix: "Report to human for decision"
+```
+
+---
+
+## Integration with AIOS
+
+This task creates squads that seamlessly integrate with:
+- Core AIOS-FULLSTACK framework
+- Standard installer (`npm run install:squad`)
+- Memory layer for tracking
+- Agent activation system (@agent-id syntax)
+- Quality frameworks for validation
+
+---
+
+_Task Version: 2.0_
+_Last Updated: 2026-02-01_
+_Lines: 650+_