skill: qe-learning-optimization
version: 1.0.0
description: >
  Evaluation suite for QE learning optimization.
  Tests pattern learning, cross-model comparison, and ReasoningBank integration.

models_to_test:
  - claude-3.5-sonnet
  - claude-3-haiku

mcp_integration:
  enabled: true
  namespace: skill-validation
  query_patterns: true
  track_outcomes: true
  store_patterns: true
  target_agents:
    - qe-learning-coordinator

learning:
  store_success_patterns: true
  store_failure_patterns: true
  pattern_ttl_days: 90

result_format:
  json_output: true
  include_timing: true
  include_token_usage: true

setup:
  required_tools:
    - jq

test_cases:

  - id: tc001_pattern_learning
    description: "Learn and store test patterns"
    category: learning
    priority: critical

    input:
      test_case:
        name: "sql_injection_detection"
        input: "SELECT * FROM users WHERE id = ${userId}"
        output: "SQL injection vulnerability detected"
        success: true

    expected_output:
      must_contain:
        - "pattern"
        - "learning"
        - "store"

    validation:
      schema_check: true
      keyword_match_threshold: 0.8

  - id: tc002_cross_model_comparison
    description: "Compare outputs across different models"
    category: comparison
    priority: high

    input:
      models:
        - "claude-3.5-sonnet"
        - "claude-3-haiku"
      test_case: "security_vulnerability"

    expected_output:
      must_contain:
        - "comparison"
        - "variance"

    validation:
      schema_check: true

  - id: tc003_pattern_query
    description: "Query and retrieve learned patterns"
    category: retrieval
    priority: high

    input:
      pattern_type: "security_finding"
      domain: "sql_injection"

    expected_output:
      must_contain:
        - "pattern"
        - "retrieve"

    validation:
      schema_check: true

  - id: tc004_confidence_scoring
    description: "Score pattern confidence and reliability"
    category: scoring
    priority: medium

    input:
      pattern:
        success_rate: 0.95
        test_count: 20
        domains: ["security", "api"]

    expected_output:
      must_contain:
        - "confidence"
        - "score"

    validation:
      schema_check: true

  - id: tc005_learning_feedback
    description: "Record learning feedback loop"
    category: feedback
    priority: medium

    input:
      test_outcome:
        passed: true
        pattern_used: "security_injection_detection"
        confidence: 0.92

    expected_output:
      must_contain:
        - "feedback"
        - "learning"

    validation:
      schema_check: true
      allow_partial: true

success_criteria:
  pass_rate: 0.9
  critical_pass_rate: 1.0
  avg_reasoning_quality: 0.75
  max_execution_time_ms: 300000

metadata:
  author: "qe-learning-coordinator"
  created: "2026-02-02"
  coverage_target: >
    QE learning optimization with 5 test cases covering pattern learning,
    cross-model comparison, pattern retrieval, confidence scoring, and feedback loops.