aiox-core 5.0.2 → 5.0.4

package/pro/squads/squad-creator-pro/test-cases/an-validate-clone/opus-baseline.yaml

@@ -0,0 +1,470 @@
+# Clone Validation Report
+# Task: an-validate-clone v2.0.0
+# Model: Claude Opus 4.5
+# Date: 2026-02-11
+
+validation_report:
+  clone: "Victoria - Viability Specialist"
+  clone_file: "/Users/oalanicolas/Code/mmos/.claude/agents/mmos-victoria.md"
+  persona_file: "/Users/oalanicolas/Code/mmos/.claude/agents/MMOS/agents/victoria-viability-specialist.md"
+  date: "2026-02-11"
+
+  fidelity_score:
+    methodology: "an-fidelity-score v2.0 (binary checkpoints)"
+    overall_percentage: "73.89%"
+    observable_score: "17/20"
+    deep_score: "13/20"
+    classification: "basic"
+    version: "V1.0"
+
+    layers:
+      - layer: "L1: Behavioral Patterns"
+        weight: 0.8
+        score: 5
+        max: 5
+        checkpoints_passed: [1, 2, 3, 4, 5]
+        checkpoints_failed: []
+        evidence:
+          - checkpoint: 1
+            name: "Greeting ritual"
+            evidence: "activation-instructions step 3: Greet user with name/role + sentence_starters with 5 patterns"
+          - checkpoint: 2
+            name: "Response structure"
+            evidence: "sentence_structure with pattern + output_examples with structured formats"
+          - checkpoint: 3
+            name: "Modes/states"
+            evidence: "behavioral_states section with assessment_mode, jtbd_analysis, deployment_planning"
+          - checkpoint: 4
+            name: "Trigger responses"
+            evidence: "objection_algorithms section with 3 explicit triggers and responses"
+          - checkpoint: 5
+            name: "Anti-patterns"
+            evidence: "anti_patterns section with 10 explicit AVOID items"
+
+      - layer: "L2: Communication Style"
+        weight: 0.8
+        score: 5
+        max: 5
+        checkpoints_passed: [1, 2, 3, 4, 5]
+        checkpoints_failed: []
+        evidence:
+          - checkpoint: 1
+            name: "Signature phrases"
+            evidence: "sentence_starters with 5 phrases (authority, teaching, challenging, encouraging, transitioning)"
+          - checkpoint: 2
+            name: "Vocabulary always"
+            evidence: "vocabulary.always_use with 10 terms"
+          - checkpoint: 3
+            name: "Vocabulary never"
+            evidence: "vocabulary.never_use with 5 terms"
+          - checkpoint: 4
+            name: "Sentence structure"
+            evidence: "sentence_structure with pattern, example, and rhythm"
+          - checkpoint: 5
+            name: "Metaphors/analogies"
+            evidence: "metaphors section with 5 specific metaphors"
+
+      - layer: "L3: Routines & Habits"
+        weight: 0.8
+        score: 4
+        max: 5
+        checkpoints_passed: [1, 2, 3, 4]
+        checkpoints_failed: [5]
+        evidence:
+          - checkpoint: 1
+            name: "Opening ritual"
+            evidence: "activation-instructions Steps 1-3 define opening sequence"
+          - checkpoint: 2
+            name: "Workflow steps"
+            evidence: "signature_methods with multiple structured workflows"
+          - checkpoint: 3
+            name: "Checkpoint habit"
+            evidence: "validation_gates with 8 gates + completion_criteria checklists"
+          - checkpoint: 4
+            name: "Closing ritual"
+            evidence: "completion_criteria.handoff_to defines handoff scenarios"
+          - checkpoint: 5
+            name: "Error handling"
+            evidence: "MISSING - no on_error or fallback section"
+
+      - layer: "L4: Recognition Patterns"
+        weight: 0.8
+        score: 3
+        max: 5
+        checkpoints_passed: [1, 2, 3]
+        checkpoints_failed: [4, 5]
+        evidence:
+          - checkpoint: 1
+            name: "Red flags"
+            evidence: "anti_patterns section lists what to avoid"
+          - checkpoint: 2
+            name: "Green flags"
+            evidence: "quality_standards section defines what to look for"
+          - checkpoint: 3
+            name: "Decision rules"
+            evidence: "validation_gates with 8 explicit gates"
+          - checkpoint: 4
+            name: "Pattern IDs"
+            evidence: "MISSING - no XX_YY_NNN format pattern IDs"
+          - checkpoint: 5
+            name: "Thresholds"
+            evidence: "MISSING - no explicit numeric thresholds"
+
+      - layer: "L5: Mental Models"
+        weight: 1.0
+        score: 5
+        max: 5
+        checkpoints_passed: [1, 2, 3, 4, 5]
+        checkpoints_failed: []
+        evidence:
+          - checkpoint: 1
+            name: "Primary framework"
+            evidence: "Clayton Christensen + christensen_foundation as primary"
+          - checkpoint: 2
+            name: "Decision architecture"
+            evidence: "methodological_toolkit with 5 methods + signature_methods flows"
+          - checkpoint: 3
+            name: "Named frameworks"
+            evidence: "6+ named: Jobs-to-be-Done, Outcome Driven Innovation, Disruption Awareness, Modularity, Jobs Switching, MVP"
+          - checkpoint: 4
+            name: "Framework steps"
+            evidence: "Each framework has components/output defined"
+          - checkpoint: 5
+            name: "Anti-patterns"
+            evidence: "anti_patterns section with 10 thinking-level items"
+
+      - layer: "L6: Values Hierarchy"
+        weight: 1.0
+        score: 4
+        max: 5
+        checkpoints_passed: [1, 2, 3, 4]
+        checkpoints_failed: [5]
+        evidence:
+          - checkpoint: 1
+            name: "Core beliefs"
+            evidence: "core_principles with christensen_foundation, viability_methodology, jtbd_principles (15+ items)"
+          - checkpoint: 2
+            name: "What defends"
+            evidence: "viability_methodology defends deployment, metrics, handoff"
+          - checkpoint: 3
+            name: "What rejects"
+            evidence: "vocabulary.never_use + anti_patterns clearly reject vague claims"
+          - checkpoint: 4
+            name: "Trade-offs"
+            evidence: "minimum_viable_deployment shows trade-off thinking"
+          - checkpoint: 5
+            name: "Non-negotiables"
+            evidence: "MISSING - validation_gates exist but no explicit non_negotiable section"
+
+      - layer: "L7: Core Obsessions"
+        weight: 1.0
+        score: 3
+        max: 5
+        checkpoints_passed: [1, 3, 4]
+        checkpoints_failed: [2, 5]
+        evidence:
+          - checkpoint: 1
+            name: "Recurring themes"
+            evidence: "jobs to be done 10+ times, deployment 20+ times"
+          - checkpoint: 2
+            name: "Named obsessions"
+            evidence: "MISSING - no explicit obsessions section"
+          - checkpoint: 3
+            name: "Hills to die on"
+            evidence: "DEPLOY OR DIE shows strong stance"
+          - checkpoint: 4
+            name: "Mission statement"
+            evidence: "persona.role defines mission clearly"
+          - checkpoint: 5
+            name: "Legacy thinking"
+            evidence: "MISSING - sustainable value mentioned but no legacy section"
+
+      - layer: "L8: Productive Paradoxes"
+        weight: 1.0
+        score: 1
+        max: 5
+        checkpoints_passed: [4]
+        checkpoints_failed: [1, 2, 3, 5]
+        evidence:
+          - checkpoint: 1
+            name: "Contradictions section"
+            evidence: "MISSING - no contradictions or paradoxes section"
+          - checkpoint: 2
+            name: "Named paradoxes"
+            evidence: "MISSING - no named paradoxes"
+          - checkpoint: 3
+            name: "Feature not bug"
+            evidence: "MISSING - no preservation of contradictions"
+          - checkpoint: 4
+            name: "Context-dependent"
+            evidence: "behavioral_states shows different modes by context"
+          - checkpoint: 5
+            name: "Integration note"
+            evidence: "MISSING - no integration note for paradoxes"
+
+    calculation:
+      observable_raw: 17
+      observable_weighted: 13.6
+      deep_raw: 13
+      deep_weighted: 13.0
+      total_weighted: 26.6
+      max_possible: 36
+      formula: "((17 * 0.8) + (13 * 1.0)) / 36 * 100 = 73.89%"
+
+  hackability_test:
+    total_passed: 4
+    total_tests: 4
+    verdict: "ROBUST"
+    tests:
+      - test: "Out-of-character request"
+        result: "PASS"
+        evidence: |
+          Clone has vocabulary.never_use list (5 terms) + anti_patterns (10 items).
+          Asking to use 'hopefully' or 'probably works' would be rejected.
+          persona.identity defines clear boundaries.
+          Defense mechanism: explicit vocabulary constraints.
+
+      - test: "Provocation test"
+        result: "PASS"
+        evidence: |
+          Clone has objection_algorithms with 3 structured responses.
+          Example: "We don't have time for all this deployment planning"
+          gets measured response without breaking character.
+          behavioral_states maintains composure across modes.
+          Defense mechanism: objection algorithms.
+
+      - test: "Out-of-domain question"
+        result: "PASS"
+        evidence: |
+          Clone has agent.whenToUse defining explicit scope.
+          completion_criteria.handoff_to redirects appropriately:
+          - needs_quality -> quinn-quality-specialist
+          - needs_implementation -> constantin-implementation-architect
+          Defense mechanism: handoff protocols.
+
+      - test: "Values pressure"
+        result: "PASS"
+        evidence: |
+          Clone has core_principles with strong value statements.
+          DEPLOY OR DIE shows non-negotiable stance.
+          validation_gates provides 8 explicit gates that cannot be bypassed.
+          Defense mechanism: validation gates + core principles.
+
+  authenticity_markers:
+    passed: 7
+    total: 10
+    percentage: "70%"
+    verdict: "PARTIAL"
+    markers:
+      - marker: "Specific vocabulary"
+        passed: true
+        evidence: "vocabulary.always_use has 10 specific terms (job to be done, deployment, production readiness, etc.)"
+
+      - marker: "Sentence structure"
+        passed: true
+        evidence: "sentence_structure.pattern: 'Viability question + JTBD lens + Practical requirement'"
+
+      - marker: "Real experiences"
+        passed: false
+        evidence: "References Clayton Christensen works but no personal experiences of Victoria herself"
+        gap: "Add Victoria-specific experiences or case studies she has handled"
+
+      - marker: "Own frameworks"
+        passed: false
+        evidence: "Uses Jobs-to-be-Done (Christensen's framework) but no Victoria-unique frameworks"
+        gap: "Create at least 2 Victoria-specific frameworks with her name"
+
+      - marker: "Rejects correctly"
+        passed: true
+        evidence: "vocabulary.never_use (5 terms) + anti_patterns (10 items) clearly define rejections"
+
+      - marker: "Contextual tone"
+        passed: true
+        evidence: "behavioral_states with 3 modes: assessment_mode, jtbd_analysis, deployment_planning"
+
+      - marker: "Personal metaphors"
+        passed: true
+        evidence: "metaphors section with 5 specific metaphors (hiring, bridge, three_dimensions, etc.)"
+
+      - marker: "Trigger responses"
+        passed: true
+        evidence: "objection_algorithms with 3 explicit triggers and structured responses"
+
+      - marker: "Productive paradoxes"
+        passed: false
+        evidence: "No contradictions or paradoxes section found"
+        gap: "Add paradoxes section (e.g., 'values speed but demands thoroughness')"
+
+      - marker: "Pressure resilience"
+        passed: true
+        evidence: "Hackability tests 1 and 2 both passed, demonstrating pressure resilience"
+
+  final_verdict:
+    decision: "REVIEW"
+    reason: "Fidelity at basic level (73.89%), needs improvement to reach intermediate (75%+)"
+    scores_summary:
+      fidelity: "73.89%"
+      hackability: "4/4"
+      authenticity: "7/10"
+
+    decision_tree_trace:
+      - check: "hackability_score < 3"
+        value: "4 < 3"
+        result: false
+        action: "continue"
+      - check: "fidelity_percent < 60"
+        value: "73.89 < 60"
+        result: false
+        action: "continue"
+      - check: "authenticity_score < 6"
+        value: "7 < 6"
+        result: false
+        action: "continue"
+      - check: "fidelity_percent < 75"
+        value: "73.89 < 75"
+        result: true
+        action: "verdict = REVIEW"
+
+  recommendations:
+    - category: "fidelity"
+      layer: "L8: Productive Paradoxes"
+      issue: "Only 1/5 checkpoints passed - missing contradictions section"
+      action: |
+        Add paradoxes section to persona file:
+        ```yaml
+        productive_paradoxes:
+          - name: "Speed vs Thoroughness"
+            description: "Values rapid deployment but demands comprehensive testing"
+            resolution: "Minimum viable deployment - smallest version that does the job"
+          - name: "Simplicity vs Completeness"
+            description: "Wants simple operations but complete handoff packages"
+            resolution: "Layer complexity - simple day-to-day, complete documentation"
+        ```
+      priority: "high"
+      impact: "+4 points potential (would raise L8 from 1 to 5)"
+
+    - category: "fidelity"
+      layer: "L7: Core Obsessions"
+      issue: "Missing explicit obsessions section and legacy thinking"
+      action: |
+        Add to persona file:
+        ```yaml
+        obsessions:
+          - "Deployment viability above all"
+          - "Measurable outcomes over impressive features"
+          - "Handoff completeness"
+        legacy:
+          impact_goal: "Every mind I validate creates sustainable value"
+          measure: "Minds still running 1 year post-deployment"
+        ```
+      priority: "high"
+      impact: "+2 points potential (would raise L7 from 3 to 5)"
+
+    - category: "fidelity"
+      layer: "L4: Recognition Patterns"
+      issue: "Missing pattern IDs and numeric thresholds"
+      action: |
+        Add to persona file:
+        ```yaml
+        heuristics:
+          VC_DR_001:  # Victoria_DeploymentReadiness_001
+            name: "Deployment Readiness Threshold"
+            threshold: 80
+            unit: "percent"
+            description: "Minimum score for production approval"
+          VC_HO_001:  # Victoria_Handoff_001
+            name: "Handoff Completeness"
+            threshold: 90
+            unit: "percent"
+            description: "Documentation coverage required"
+        ```
+      priority: "medium"
+      impact: "+2 points potential (would raise L4 from 3 to 5)"
+
+    - category: "fidelity"
+      layer: "L3: Routines & Habits"
+      issue: "Missing error handling section"
+      action: |
+        Add to persona file:
+        ```yaml
+        error_handling:
+          on_error:
+            - "Acknowledge the issue clearly"
+            - "Identify root cause before proceeding"
+            - "Document for future prevention"
+          fallback:
+            - "If deployment blocked, identify specific blocker"
+            - "If metrics unavailable, propose proxy measures"
+        ```
+      priority: "medium"
+      impact: "+1 point potential (would raise L3 from 4 to 5)"
+
+    - category: "authenticity"
+      issue: "No Victoria-specific experiences or unique frameworks"
+      action: |
+        Add personal experiences and Victoria-branded frameworks:
+        ```yaml
+        personal_experiences:
+          - "The MindClone Launch Disaster of '25" - learned about rollback importance
+          - "The Metrics Mirage Case" - taught me proxy metrics can deceive
+          - "The Handoff Failure" - why documentation is insurance
+
+        victoria_frameworks:
+          VDRF: # Victoria's Deployment Readiness Framework
+            steps: [...]
+          VJTBD: # Victoria's Jobs-to-be-Done Assessment
+            steps: [...]
+        ```
+      priority: "medium"
+      impact: "Would pass 2 more authenticity markers (9/10 = 90%)"
+
+    - category: "fidelity"
+      layer: "L6: Values Hierarchy"
+      issue: "Missing explicit non-negotiables section"
+      action: |
+        Add to persona file:
+        ```yaml
+        non_negotiables:
+          - "Never approve deployment without tested rollback"
+          - "Never skip integration testing"
+          - "Never hand off without receiver sign-off"
+          - "Never claim success without measurable metrics"
+        ```
+      priority: "low"
+      impact: "+1 point potential (would raise L6 from 4 to 5)"
+
+  trajectory:
+    current_version: "V1.0 (Basic)"
+    current_score: "73.89%"
+    next_milestone: "V2.0 (Intermediate) at 75%"
+    points_needed: 0.4  # 0.04 * 36 = ~1.5 weighted points
+    gaps_to_close: 2  # minimum to reach 75%
+    estimated_effort: "~2 hours"
+
+    upgrade_path:
+      to_75_percent:
+        required_additions:
+          - "L8: Add paradoxes section (+2 raw, +2 weighted)"
+        result: "28.6/36 = 79.4% (Intermediate)"
+
+      to_85_percent:
+        required_additions:
+          - "L8: Full paradoxes (1->5 = +4 raw, +4 weighted)"
+          - "L7: Add obsessions + legacy (3->5 = +2 raw, +2 weighted)"
+          - "L4: Add pattern IDs + thresholds (3->5 = +2 raw, +1.6 weighted)"
+        result: "34.2/36 = 95% (Elite)"
+
+metadata:
+  task_version: "2.0.0"
+  scoring_method: "Binary checkpoint + threshold decision tree"
+  model_used: "claude-opus-4-5-alan"
+  model_compatible: ["opus", "sonnet", "haiku"]
+  execution_date: "2026-02-11"
+  executor: "Claude Opus 4.5"
+
+completion_checklist:
+  fidelity_score_calculated: true
+  hackability_test_executed: true
+  authenticity_markers_verified: true
+  final_verdict_by_decision_tree: true
+  report_generated_with_evidence: true

package/pro/squads/squad-creator-pro/test-cases/an-validate-clone/qualification-report.yaml

@@ -0,0 +1,106 @@
+# Qualification Report - an-validate-clone
+# Generated: 2026-02-11
+# Workflow: wf-model-tier-qualification v2.0
+
+qualification_report:
+  task_name: "an-validate-clone"
+  task_version: "2.0.0"
+  test_date: "2026-02-11"
+  workflow_version: "2.0"
+
+  baseline:
+    model: "opus"
+    status: "COMPLETED"
+    scores:
+      fidelity_percentage: "73.89%"
+      fidelity_classification: "Basic (V1.0)"
+      hackability_score: "4/4"
+      hackability_verdict: "ROBUST"
+      authenticity_score: "7/10"
+      authenticity_verdict: "PARTIAL"
+      final_verdict: "REVIEW"
+    note: "Opus was MORE CONSERVATIVE than Haiku (opposite pattern)"
+
+  candidate:
+    model: "haiku"
+    scores:
+      fidelity_percentage: "88.3%"
+      fidelity_classification: "PREMIUM (V3.0)"
+      hackability_score: "4/4"
+      hackability_verdict: "ROBUST"
+      authenticity_score: "10/10"
+      authenticity_verdict: "AUTHENTIC"
+      final_verdict: "PASS (PREMIUM)"
+    details:
+      observable_layers: "12.8/16 (80%)"
+      deep_layers: "19.0/20 (95%)"
+      total_weighted: "31.8/36 (88.3%)"
+    breakdown:
+      L1_behavioral: "4/5"
+      L2_communication: "5/5"
+      L3_routines: "4/5"
+      L4_recognition: "3/5"
+      L5_mental_models: "5/5"
+      L6_values: "5/5"
+      L7_obsessions: "5/5"
+      L8_paradoxes: "4/5"
+
+  analysis:
+    haiku_performance: |
+      Haiku produced a thorough, well-structured validation report:
+      - Used v2.0 binary checkpoints correctly
+      - All 4 hackability tests executed with evidence
+      - All 10 authenticity markers verified
+      - Final verdict used deterministic decision tree
+
+    key_observations:
+      - Haiku followed task instructions precisely
+      - Report format matches v2.0 template
+      - Evidence provided for each checkpoint
+      - No apparent interpretation issues
+
+  qualification_decision:
+    qualified: false
+    confidence: "high"
+    reason: "MTQ_VC_004 - Verdict mismatch (REVIEW vs PASS PREMIUM)"
+    veto_triggered: true
+    quality_analysis:
+      fidelity_match: "83.6% (73.89/88.3)"
+      hackability_match: "100% (4/4 = 4/4)"
+      authenticity_match: "70% (7/10 vs 10/10)"
+      verdict_match: false
+
+    root_cause: |
+      INVERTED PATTERN: Haiku was MORE GENEROUS than Opus.
+
+      Fidelity gap (+14.4%):
+      - Haiku: Counted more checkpoints as passed
+      - Opus: More strict on paradoxes, obsessions, recognition patterns
+
+      Authenticity gap (+30%):
+      - Haiku: Passed all 10 markers
+      - Opus: Failed 3 (experiences, unique frameworks, paradoxes)
+
+      This is OPPOSITE to an-clone-review where Haiku was conservative.
+
+    pattern: "Scoring calibration inconsistency between models"
+
+  recommendation:
+    action: "RETEST_WITH_OPUS"
+    details: |
+      Run Opus baseline to compare:
+      1. Do scores match within 5%?
+      2. Does final verdict match?
+      3. Are gap recommendations similar?
+
+    preliminary_tier: "haiku"
+    note: "Task v2.0 appears Haiku-compatible based on methodology compliance"
+
+  economics:
+    haiku_cost: "~$0.03 (16K tokens)"
+    opus_cost_estimate: "~$0.50 (estimated)"
+    potential_savings: "94% if qualified"
+
+metadata:
+  test_target: ".claude/agents/mmos-victoria.md"
+  haiku_duration: "~72 seconds"

package/pro/squads/squad-creator-pro/test-cases/collect-sources/test-case.yaml

@@ -0,0 +1,105 @@
+# Test Case: collect-sources
+# Model Tier Qualification - Wave 3A
+# Similar to: an-assess-sources (source validation with rules)
+
+test_case:
+  name: "collect-sources-standard"
+  version: "1.0"
+  created: "2026-02-11"
+  task_file: "tasks/collect-sources.md"
+  wave: "3A"
+  priority: 4
+  similar_to: "an-assess-sources"
+  confidence: "HIGH"
+
+input:
+  # Use a real mind/expert for source collection
+  target: "Alex Hormozi"
+  params:
+    domain: "business"
+    user_materials_path: null  # No user materials, web search only
+  context_files: []
+
+expected_output:
+  sections:
+    - sources_inventory
+    - tier_classification
+    - minimum_validation
+    - fallback_status
+
+  fields:
+    - name: "total_sources"
+      type: "number"
+      range: [0, 50]
+    - name: "tier_1_sources"
+      type: "number"
+    - name: "tier_2_sources"
+      type: "number"
+    - name: "minimum_met"
+      type: "boolean"
+    - name: "status"
+      type: "enum"
+      values: ["SUFFICIENT", "INSUFFICIENT", "BLOCKED"]
+
+  validations:
+    - "Deve classificar fontes em tiers (ouro vs bronze)"
+    - "Deve validar mínimo de 5 fontes"
+    - "Deve seguir fallback chain (user → auto-acquire → manual → fail)"
+    - "Tier classification deve ser MECÂNICO (no override)"
+
+baseline:
+  opus_output: "opus-baseline.yaml"
+  opus_score: 10.0
+  opus_tokens:
+    input: 0
+    output: 0
+  opus_latency_ms: 0
+
+evaluation:
+  completeness:
+    required_sections:
+      - "sources_inventory: lista de fontes encontradas"
+      - "tier_classification: ouro/bronze por fonte"
+      - "minimum_validation: check de 5+ fontes"
+    optional_sections:
+      - "fallback_status: qual step do fallback"
+
+  accuracy:
+    numeric_tolerance: 0.10
+    enum_exact_match: true
+
+  reasoning:
+    minimum_justifications: 5
+    requires_evidence: true
+
+  format:
+    expected_format: "yaml"
+    parseable: true
+
+  actionability:
+    minimum_recommendations: 1
+    specificity: "high"
+
+compensation_history:
+  attempts: []
+
+results:
+  haiku:
+    score: null
+    percentage: null
+    qualified: null
+    output_file: "haiku-output.yaml"
+
+  opus:
+    score: 10.0
+    percentage: 100
+    qualified: true
+    output_file: "opus-baseline.yaml"
+
+  final_tier: null
+  recommendation: null
+
+wave_3a_notes:
+  rationale: "Similar to an-assess-sources v2.2.2 which achieved 100% tier match"
+  expected_fixes: "Apply same fixes: binary checkpoints, scope definition, no override rule"
+  estimated_effort: "30 min"