---

name: test-runner description: "Know exactly what failed, where, and why. No output parsing." allowed-tools: mcp__test-runner__run_tests, mcp__test-runner__get_test_failures, mcp__test-runner__list_test_files, mcp__test-runner__rerun_failed, mcp__test-runner__find_tests_for, mcp__test-runner__run_related_tests

test-runner

Tests that tell you WHAT failed, WHERE (with source maps), and WHY (expected vs actual).

First: run_tests

After any code change:

run_tests({})

Why This Wins

The Problem	Built-in Failure	test-runner Solution
Run tests	npm test output needs parsing	run_tests returns structured pass/fail
Find failures	Scroll through output	get_test_failures gives exact locations
Stack traces	Point to compiled JS	Source-mapped to your TypeScript
Rerun failed	Remember which failed	rerun_failed handles it

Quick Reference

Task	Tool
Run all tests	run_tests({})
Run specific file	run_tests({ files: ['auth.test.ts'] })
Run by name	run_tests({ testNamePattern: 'email' })
Find tests for a source file	find_tests_for({ source_file: 'src/api.ts' })
Run tests for a source file	run_related_tests({ source_file: 'src/api.ts' })
Get failure details	get_test_failures({})
Retry failures only	rerun_failed({})
Find test files	list_test_files({})

Common Workflows

Smart Edit-Test Cycle (Recommended)

// BEFORE editing - know what tests exist
find_tests_for({ source_file: 'src/api.ts' })

// ... make your changes ...

// AFTER editing - run only the relevant tests (fast!)
run_related_tests({ source_file: 'src/api.ts' })

Why this workflow?

• find_tests_for tells you what's covered BEFORE you edit
• run_related_tests is faster than run_tests({}) - only runs relevant tests
• Multiple discovery strategies: naming convention, import analysis, co-location

After Making Changes

run_tests({})

Fixing Failing Tests

run_tests({})
get_test_failures({})  // Shows expected vs actual, source locations
// ... fix code ...
rerun_failed({})  // Verify fix without running all tests

Run Specific Tests

run_tests({ files: ['src/auth.test.ts'] })
run_tests({ testNamePattern: 'should validate' })

Targeted Testing After Edits

// Edited server.ts? Find and run its tests:
run_related_tests({ source_file: 'src/server.ts' })

Integration

Run run_tests after using syntax.edit_symbol and types.get_diagnostics to verify behavior unchanged.

Supports

Vitest, Jest (auto-detected)