Test Tier Selection

Automatic guidance for selecting appropriate test tiers based on change context and scope.

Test Tier Definitions

| Tier | Duration | Scope | When to Run | |------|----------|-------|-------------| | Unit | < 30s | Single function/module | After every code change | | Integration | < 5min | Component interactions | After feature completion | | E2E | < 30min | Full user flows | Before commit/PR |

Decision Matrix

Change Type → Test Tier

| Change Type | Unit | Integration | E2E | |-------------|------|-------------|-----| | Single function fix | Required | Skip | Skip | | New feature (1 file) | Required | Required | Skip | | Multi-file feature | Required | Required | Required | | Refactoring | Required | Required | Optional | | API changes | Required | Required | Required | | UI changes | Required | Optional | Required | | Bug fix (isolated) | Required | Optional | Skip | | Database changes | Required | Required | Required | | Config changes | Required | Required | Optional |

Escalation Signals

Escalate to Integration when:

• Changes span multiple files
• Business logic affected
• Service boundaries modified
• Database queries changed

Escalate to E2E when:

• User-facing features modified
• Authentication/authorization changes
• Critical path functionality
• Before creating PR

Commands by Tier

# Tier 1: Unit (fast feedback)
/test:quick

# Tier 2: Integration (feature completion)
/test:full --coverage

# Tier 3: E2E (pre-commit)
/test:full

Agent Consultation Triggers

Consult test-architecture agent when:

• New feature module created
• Coverage drops > 5%

• 3 flaky tests detected

• Framework questions arise
• Test strategy needs adjustment

Consult test-runner agent when:

• Need test execution with analysis
• Multiple failures to diagnose
• Want concise failure summary

Consult system-debugging agent when:

• Integration test failures with unclear cause
• Environment/timing issues
• Flaky tests related to concurrency

Quick Reference

After Small Change

1. Run /test:quick
2. If pass: Continue working
3. If fail: Fix immediately

After Feature Completion

1. Run /test:full --coverage
2. Check coverage targets met
3. If gaps: /test:consult coverage

Before Commit/PR

1. Run /test:full
2. All tiers must pass
3. Review coverage report

For New Features

1. /test:consult new-feature
2. Write tests (TDD)
3. Run /test:quick during development
4. Run /test:full before PR

Activation Triggers

This skill auto-activates when:

• User mentions "test", "run tests", "testing"
• After code modification by Claude
• During TDD workflow
• When /test:* commands invoked
• When discussing test strategy