testing-theatre-detection

name	testing-theatre-detection
description	Detect and eliminate testing theatre - tests that can't prevent regressions

Test Quality Auditor

Purpose

Enforce zero tolerance for testing theatre across all test types.

MANDATORY: Structured User Input

Whenever soliciting user input — choices, confirmations, or selections — ALWAYS use the AskUserQuestion tool. NEVER ask via prose. Follow the format in .claude/rules/ask-user-question-format.md: Re-ground context, Predict, Options.

When to Use

1. Manual: User runs /testing-theatre-detection
2. Automatic: Invoked by code-review skill

Process

Announce professional mode status:

Using testing-theatre-detection skill. Professional mode is ACTIVE - architect mode enforced (no code changes).

Determining Scope

If invoked with exact file path: Skip scope determination. Analyze the provided file directly. Example: /testing-theatre-detection src/auth/login.test.js

If invoked from plan execution (subagent with task context): Identify tests related to the specific task/feature:

1. Check task description for mentioned files/components
2. Use Grep to find test files importing those components
3. Use git diff to find modified test files in task branch
4. Analyze all identified test files

If manual invocation (no file path provided): Use AskUserQuestion tool:

• question: "What scope should I analyze?"
• header: "Analysis scope"
• options: "Current changes" (test files in git diff) | "Entire test suite" (all test files in the project) | "Specific path" (I'll specify a path) | "Task-specific" (tests for a particular feature or task)

If auto-invoked (from code-review): Automatically use current changes (git diff --name-only)

Finding Test Files

Run appropriate command based on scope:

• Single file: Use the exact file path provided (skip discovery)
• Task-specific:
  • Parse task description for component/file names
  • Use Grep to find test files: import.*{ComponentName} or import.*'path/to/file'
  • Use git diff if on task branch
• Current changes: git diff --name-only | grep -E '\.(test|spec)\.(js|ts|tsx|java|py)$'
• Entire suite: Use Glob tool with patterns: **/*.test.js, **/*Test.java, **/test_*.py
• Specific path: Use Glob tool with user-provided path

Auto-Detecting Test Framework

For each test file, determine framework:

Jest Detection:

• File extension: .test.js, .spec.js, .test.ts, .spec.ts, .test.tsx, .spec.tsx
• Import patterns: import { test, expect } from, import { describe, it } from
• Framework: jest

JUnit Detection:

• File pattern: *Test.java, *Tests.java
• Import patterns: import org.junit, @Test, @Disabled
• Framework: junit

pytest Detection:

• File pattern: test_*.py, *_test.py
• Import patterns: import pytest, @pytest.mark
• Framework: pytest

React Testing Library Detection:

• File extension: .test.tsx, .spec.tsx
• Import patterns: import { render } from '@testing-library/react'
• Framework: jest + react-testing-library

If framework cannot be determined:

• Use AskUserQuestion tool:
  • question: "What test framework is {file} using?"
  • header: "Test framework"
  • options: "Jest" (JavaScript/TypeScript) | "JUnit" (Java) | "pytest" (Python) | "React Testing Library" (Jest + RTL)
• Fall back to generic pattern matching

Static Analysis Phase

Check 1: Pending/Skipped Tests

Jest patterns to detect:

• it.skip(, test.skip(, xit(, xtest(, describe.skip(
• Search pattern: Use Grep tool with: \.(skip|xit|xtest)\(

JUnit patterns to detect:

• @Disabled, @Ignore annotations
• Search pattern: Use Grep tool with: @(Disabled|Ignore)

pytest patterns to detect:

• @pytest.mark.skip, @pytest.mark.xfail
• Search pattern: Use Grep tool with: @pytest\.mark\.(skip|xfail)

For each match found:

1. Extract line number
2. Extract test name from surrounding context
3. Add to issues list:

   Issue: Skipped Test (Critical)
   Line {number}: {test name}
   Problem: Test is disabled - known broken behavior being ignored
   Risk: Production bug hiding behind disabled test
   Fix: Remove skip/disable annotation and fix the failing test
   

Check 2: Missing/Weak Assertions

Detection approach: For each test function, count assertion statements.

Jest assertions to count:

• expect( statements
• assert( statements
• Search pattern: Use Grep tool with: expect\(|assert\(

JUnit assertions to count:

• assert, assertEquals, assertTrue, assertFalse, etc.
• Search pattern: Use Grep tool with: assert[A-Z]\w+\(

pytest assertions to count:

• assert statements
• Search pattern: Use Grep tool with: ^\s+assert\s

Tautological assertion patterns:

• Jest: expect(x).toBe(x), expect(true).toBe(true)
• JUnit: assertTrue(true), assertEquals(x, x)
• Python: assert True, assert x == x

For each test with zero assertions:

1. Extract test name and line number
2. Add to issues list with code example:

   Issue: No Assertions (Critical)
   Line {number}: test "{name}"
   
   Problem: Test has no expect() calls - always passes regardless of implementation
   Risk: Cannot detect regressions in {component} behavior
   
   Fix:
   test('{name}', () => {
     // Arrange
     const result = functionUnderTest(input);
   
     // Assert
     expect(result).toBe(expectedValue);
     expect(result.property).toEqual(expectedProperty);
   });
   

Check 3: Over-Mocking

Detection approach: Count mock statements vs real code invocations in each test.

Mock patterns to count:

Jest:

• jest.mock(, jest.spyOn(, mockImplementation, mockReturnValue
• Search pattern: jest\.(mock|spyOn)|mock(Implementation|ReturnValue)

JUnit:

• @Mock, Mockito.mock(, when(, verify(
• Search pattern: @Mock|Mockito\.(mock|when|verify)

pytest:

• @patch, Mock(), MagicMock()
• Search pattern: @patch|Mock\(\)|MagicMock\(\)

Calculate ratio:

• Mock lines / total test lines
• If ratio > 0.8 (80%), flag as over-mocking

For each over-mocked test:

1. Add to issues list:

   Issue: Over-Mocking (Critical)
   Line {number}: test "{name}"
   
   Problem: {percentage}% of test is mocking - not testing real behavior
   Risk: Tests pass but production code may be broken
   
   Fix: Reduce mocking. Test real integrations when possible:
   - Mock external dependencies (APIs, databases) at boundaries
   - Use real implementations for internal logic
   - Integration tests should test actual integration
   

Check 4: Snapshot-Only Tests

Detection approach: Find tests that ONLY use snapshot assertions without behavioral assertions.

Snapshot patterns:

• Jest: toMatchSnapshot(), toMatchInlineSnapshot()
• Search pattern: toMatchSnapshot|toMatchInlineSnapshot

Check logic:

1. Count snapshot assertions in test
2. Count non-snapshot assertions (expect, assert, etc.)
3. If snapshots > 0 AND non-snapshot === 0, flag as snapshot-only

For each snapshot-only test:

1. Add to issues list:

   Issue: Snapshot Only (Critical)
   Line {number}: test "{name}"
   
   Problem: Only uses toMatchSnapshot() with no behavior validation
   Risk: Doesn't verify {component} actually works
   
   Fix: Add assertions for interactive behavior:
   - Test click handlers are called with correct arguments
   - Test state changes correctly
   - Test props are applied correctly
   - Test accessibility attributes
   - Then use snapshots as supplementary check
   

Check 5: Error Swallowing

Detection approach: Find try/catch blocks or conditional logic that can prevent test failures.

Patterns to detect:

Try/catch with no rethrow:

• Jest/JS: try { ... } catch (e) { } or catch (e) { console.log }
• Java: catch (Exception e) { } with no throw/fail
• Python: except: pass or except Exception:

Conditional assertions:

• if (condition) { expect(...) } - assertion might not run

Search patterns:

• Try/catch: Use Grep with: catch\s*\([^)]+\)\s*\{\s*\}
• Conditional assertions: Use Read tool to parse test structure

For each error swallowing pattern:

1. Add to issues list:

   Issue: Error Swallowing (Critical)
   Line {number}: test "{name}"
   
   Problem: Try/catch or conditional logic can prevent test failure
   Risk: Test passes even when code throws errors
   
   Fix: Either:
   - Remove try/catch and let test fail on error
   - If testing error handling, assert the error: expect(() => fn()).toThrow()
   - Remove conditional logic around assertions
   

Dynamic Analysis Phase

Running Tests

Find test command:

1. Check package.json (for JS/TS projects):

   • Read file with Read tool
   • Parse JSON and look for scripts.test or scripts.test-ci
   • Command: npm test or yarn test

2. Check Makefile (for any project):

   • Use Grep to find: ^test:|^test-.*:
   • Command: make test

3. Check build.gradle (for Java projects):

   • Use Grep to find: task test
   • Command: ./gradlew test

4. Direct framework invocation (fallback):

   • Jest: npx jest --coverage
   • JUnit: ./gradlew test or mvn test
   • pytest: pytest --cov

5. If ambiguous, use AskUserQuestion tool:

   • question: "I found multiple test commands. Which should I use?"
   • header: "Test command"
   • options: one option per discovered command (label = command string, description = where it was found)

Execute test command: Run with Bash tool:

• Capture stdout, stderr, and exit code
• Set timeout to 10 minutes (tests can be slow)
• Continue even if exit code != 0

Example:

npm test 2>&1

Parsing Test Output

Check 1: Exit Code

• If exit code != 0: Tests failed
• Add to issues:

  ❌ Test Failures (Critical)
  
  Problem: {count} tests failed
  Risk: Cannot assess test quality when tests don't pass
  
  Fix: Address test failures first:
  {list of failed tests from output}
  

Check 2: Warning Messages

Scan output for warning patterns:

• WARN:, WARNING:, Warning:
• deprecated, deprecation
• (node:) with warning

For each warning:

⚠️ Test Warning (Critical)

Problem: Test output contains warning
Warning: {warning text}
Risk: Warnings indicate unreliable test behavior

Fix: Address the warning - update deprecated APIs, fix configuration

Check 3: Error Messages (even if tests pass)

Scan output for error patterns:

• Error:, ERROR:
• Exception:
• failed to

For each error in passing tests:

❌ Hidden Error (Critical)

Problem: Test output contains errors but tests still pass
Error: {error text}
Risk: Test is swallowing errors

Fix: Update test to fail on errors or fix the underlying issue

Check 4: Flaky Indicators

Scan for flaky patterns:

• timeout, timed out
• ETIMEDOUT, ECONNREFUSED
• UnhandledPromiseRejection
• race condition

For each flaky indicator:

🔀 Flaky Test Indicator (Critical)

Problem: Test output suggests flaky/unreliable behavior
Indicator: {text}
Risk: Test may pass/fail randomly

Fix:
- Add proper async/await handling
- Increase timeouts if necessary
- Fix race conditions
- Mock unstable dependencies

Coverage Analysis

Find coverage reports:

Jest (JSON format):

• Default location: coverage/coverage-final.json
• Read with Read tool
• Parse JSON to get coverage metrics

JUnit (JaCoCo XML format):

• Default location: build/reports/jacoco/test/jacocoTestReport.xml
• Read with Read tool
• Parse XML for coverage percentages

pytest (JSON format via pytest-cov):

• Default location: .coverage or coverage.json
• Read with Read tool

Coverage metrics to extract:

• Statements covered / total
• Branches covered / total
• Functions covered / total
• Lines covered / total

Correlation with static analysis:

1. Compare coverage % to assertion count from static analysis
2. Flag: High coverage (>80%) with low assertion count (<10)

⚠️ High Coverage, Low Assertions (Critical)

Problem: {coverage}% code coverage but only {count} assertions
Risk: Executing code without verifying behavior - false confidence

Fix: Add meaningful assertions that verify:
- Return values
- State changes
- Side effects
- Error conditions

Generating Report

Report structure:

Testing Theatre Audit Report
=============================

Scope: {scope description} ({count} test files analyzed)
Status: {✅ CLEAN or ❌ {count} issues found (MUST FIX)}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📁 {file_path} ({issue_count} issues)

  {for each issue in file:}
  ❌ {Issue Type} (Critical)
  Line {number}: {test name}

  Problem: {description}
  Risk: {impact}

  Fix: {guidance with code example if applicable}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Summary: {total_issues} critical issues across {file_count} files
All issues must be fixed before production readiness.

Grouping logic:

1. Group issues by file
2. Within each file, sort by line number
3. Show file path relative to project root
4. Include issue count per file

Issue formatting:

• All issues labeled "Critical"
• Include line number for navigation
• Include test name for context
• Provide problem description, risk, and fix guidance
• Add code examples for concrete fixes (no assertions, skipped tests)
• Add bullet points for conceptual fixes (over-mocking, architecture)

Signal Completion

After generating the report (regardless of whether issues were found), signal completion to the workflow state manager:

Use MCP tool: mcp__plugin_ironclaude_state-manager__set_testing_theatre_checked
(no parameters required)

This sets testing_theatre_checked=1 in the database for the current session. The code-review skill reads this flag at Step 6.5 to determine whether testing-theatre-detection was invoked before assigning a final grade. Without this call, code-review caps the grade at C and the worker cannot advance.

If the MCP tool returns an error, display the error message — do NOT silently proceed.