Agentic-qe risk-based-testing

Focus testing effort on highest-risk areas using risk assessment and prioritization. Use when planning test strategy, allocating testing resources, or making coverage decisions.

install
source · Clone the upstream repo
git clone https://github.com/proffesor-for-testing/agentic-qe
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/proffesor-for-testing/agentic-qe "$T" && mkdir -p ~/.claude/skills && cp -r "$T/assets/skills/risk-based-testing" ~/.claude/skills/proffesor-for-testing-agentic-qe-risk-based-testing-369447 && rm -rf "$T"
manifest: assets/skills/risk-based-testing/SKILL.md
source content

Risk-Based Testing

<default_to_action> When planning tests or allocating testing resources:

  1. IDENTIFY risks per component (use 1-5 scale for probability and impact)
  2. PRIORITIZE: Critical (20+) → High (12-19) → Medium (6-11) → Low (1-5)
  3. ALLOCATE effort: 60% critical, 25% high, 10% medium, 5% low
  4. REASSESS continuously: Production incidents raise risk; stable code lowers it </default_to_action>

Quick Reference Card

When to Use

  • Planning sprint/release test strategy
  • Deciding what to automate first
  • Allocating limited testing time
  • Justifying test coverage decisions

Effort Allocation by Risk Score

ScorePriorityEffortAction
20-25Critical60%Comprehensive testing, multiple techniques
12-19High25%Thorough testing, automation priority
6-11Medium10%Standard testing, basic automation
1-5Low5%Smoke test, exploratory only

Apply Test Depth by Risk

await Task("Risk-Based Test Generation", {
  critical: {
    features: ['checkout', 'payment'],
    depth: 'comprehensive',
    techniques: ['unit', 'integration', 'e2e', 'performance', 'security']
  },
  high: {
    features: ['auth', 'user-profile'],
    depth: 'thorough',
    techniques: ['unit', 'integration', 'e2e']
  },
  medium: {
    features: ['search', 'notifications'],
    depth: 'standard',
    techniques: ['unit', 'integration']
  },
  low: {
    features: ['admin-panel', 'settings'],
    depth: 'smoke',
    techniques: ['smoke-tests']
  }
}, "qe-test-generator");

Step 3: Reassess Dynamically

// Production incident increases risk
await Task("Update Risk Score", {
  feature: 'search',
  event: 'production-incident',
  previousRisk: 9,
  newProbability: 5,  // Increased due to incident
  newRisk: 15         // Now HIGH priority
}, "qe-regression-risk-analyzer");

ML-Enhanced Risk Analysis

// Agent predicts risk using historical data
const riskAnalysis = await Task("ML Risk Analysis", {
  codeChanges: changedFiles,
  historicalBugs: bugDatabase,
  prediction: {
    model: 'gradient-boosting',
    factors: ['complexity', 'change-frequency', 'author-experience', 'file-age']
  }
}, "qe-regression-risk-analyzer");

// Output: 95% accuracy risk prediction per file

Agent Coordination Hints

Memory Namespace

aqe/risk-based/
├── risk-scores/*        - Current risk assessments
├── historical-bugs/*    - Bug patterns by area
├── production-data/*    - Incident data for risk
└── coverage-map/*       - Test depth by risk level

Fleet Coordination

const riskFleet = await FleetManager.coordinate({
  strategy: 'risk-based-testing',
  agents: [
    'qe-regression-risk-analyzer',  // Risk scoring
    'qe-test-generator',            // Risk-appropriate tests
    'qe-production-intelligence',   // Production feedback
    'qe-quality-gate'               // Risk-based gates
  ],
  topology: 'sequential'
});

Integration with CI/CD

# Risk-based test selection in pipeline
- name: Risk Analysis
  run: aqe risk-analyze --changes ${{ github.event.pull_request.files }}

- name: Run Critical Tests
  if: risk.critical > 0
  run: npm run test:critical

- name: Run High Tests
  if: risk.high > 0
  run: npm run test:high

- name: Skip Low Risk
  if: risk.low_only
  run: npm run test:smoke

Related Skills


Remember

With Agents: Agents calculate risk using ML on historical data, select risk-appropriate tests, and adjust scores from production feedback. Use agents to maintain dynamic risk profiles at scale.