Test Pruning Advisor

Skill for identifying tests that should be considered for removal based on the project's testing guidelines.

Purpose

Analyze test files to identify redundant, low-value, or obsolete tests that may be candidates for deletion. This skill follows the project's testing philosophy which prioritizes golden file tests over function-level unit tests.

Deletion Consideration Criteria

Based on project testing guidelines (CLAUDE.md):

1. •Golden file test replaceable unit tests: Function-level unit tests for code analysis, schema generation, or code generation logic that could be covered by testdata golden file tests
2. •Trivial tests: Tests for simple getters/setters, direct value passthrough, or obvious behavior that provides minimal value
3. •Implementation-coupled tests: Tests that are tightly coupled to internal implementation details and would break with reasonable refactoring
4. •Orphaned tests: Tests for code or features that no longer exist in the codebase

Workflow

Step 1: Select Review Target

When the review target is not explicitly specified, use AskUserQuestion to present the following options:

1. •Branch/PR changes: Review only test files added/modified in current branch vs main
2. •All tests: Review all *.test.ts files in the codebase
3. •Specific pattern: Review tests matching a specific path or pattern

Getting Test Files by Target

# Branch/PR changes - test files added or modified
git diff main --name-only -- '**/*.test.ts' | grep -v testdata

# All test files (excluding testdata golden tests which are reviewed by golden-test-reviewer)
find packages -name "*.test.ts" -not -path "*/testdata/*" -not -path "*/__generated__/*"

# Specific pattern
find packages -name "*.test.ts" -path "*/<pattern>/*" -not -path "*/testdata/*"

Step 2: Categorize Test Files

Group test files by their purpose:

Step 3: Launch Subagents for Analysis

For each test file (or batch of related files), launch a subagent using the Task tool with subagent_type: "general-purpose".

IMPORTANT: Launch all subagents in a single response with multiple Task tool calls to enable parallel execution.

Subagent Prompt

Read the prompt template from references/subagent-prompt.md and use it for each subagent.

Replace placeholders:

• •{test-file-path} - path to the test file
• •{category} - category from the table above
• •{related-source-files} - source files that the tests cover

Step 4: Aggregate Results and Generate Report

After all subagents complete:

1. •Parse JSON results from each subagent
2. •Generate summary table showing all test files:

## Review Summary

| Test File | Category | Tests | Deletable | Reason |
|-----------|----------|-------|-----------|--------|
| type-extractor/scanner.test.ts | Type extraction | 15 | 3 | Golden test overlap |
| shared/utils.test.ts | Utilities | 8 | 0 | - |

1. •List deletion candidates by reason:

## Deletion Candidates

### Golden File Test Replaceable (High Confidence)
- **file.test.ts**: `describe("feature X")` - Covered by testdata/feature-x-basic/
  - Specific tests: `it("should handle...")`, `it("should convert...")`

### Trivial Tests (Medium Confidence)
- **file.test.ts**: `it("should return input")` - Tests obvious passthrough behavior

### Implementation-Coupled (Review Recommended)
- **file.test.ts**: `it("internal state...")` - Depends on internal implementation detail

### Orphaned Tests (High Confidence)
- **file.test.ts**: `describe("OldFeature")` - OldFeature no longer exists

1. •

   Provide confidence levels:

   • •High: Clear candidates for deletion
   • •Medium: Likely candidates but verify before deleting
   • •Low: Potentially removable but keep for now
2. •

   Summary statistics:

## Summary

- Total test files analyzed: X
- Total test cases: Y
- Deletion candidates: Z
  - High confidence: A
  - Medium confidence: B
  - Low confidence: C
- Estimated test reduction: Z/Y (XX%)

Evaluation Confidence Levels

• •High Confidence: Tests that clearly meet deletion criteria with no ambiguity
• •Medium Confidence: Tests that likely meet criteria but should be verified
• •Low Confidence: Tests that might be deletable but have some value

Important Notes

• •This skill does NOT evaluate golden file test cases in testdata/ - use golden-test-reviewer for that
• •Focus on identifying tests that provide little value relative to maintenance cost
• •Consider the project's "prefer golden file tests over unit tests" guideline
• •Always provide reasoning for each deletion recommendation

Additional Resources

Reference Files

• •references/subagent-prompt.md - Prompt template for subagent analysis
• •references/evaluation-criteria.md - Detailed evaluation criteria with examples