mutation-testing_skill

Overview

This skill performs AI-powered mutation testing to evaluate and improve unit test quality for TypeScript projects. It generates focused code mutants guided by risk factors, runs targeted tests to find surviving mutants, and strengthens or creates tests to kill those survivors. Use it against a file path, directory, or the default set of changed files from git diff.

How this skill works

The skill identifies target source files, gathers context (source, tests, coverage, git history, and risk factors), and detects the project test runner. It creates an experimental branch, generates 3–5 realistic, buildable mutants per file, and validates each with a build check and test run. Surviving, non-equivalent mutants trigger automated test strengthening or new test generation, validated by running tests on both original and mutated code before finalizing changes and producing a mutation testing report.

When to use it

• Assess test completeness for recently changed code (default: git diff changed files)
• Validate critical code paths (security, integrations, database, performance)
• Before a release to find fragile or missing assertions
• As part of QA to reduce regression risk in shared modules or utilities
• When improving test suites to increase confidence in refactors and deployments

Best practices

• Limit mutants to 3–5 per file and keep each change atomic and realistic
• Run a build/typecheck after every mutant to avoid untestable mutations
• Prioritize risk factors (security, integrations, historic defects) when selecting mutation operators
• Never alter tests just to make mutants pass; strengthen assertions or create targeted new tests
• Keep mutant commits on an experimental branch and revert mutants before merging test improvements

Example use cases

• Harden authentication logic tests by generating security-related mutants that remove auth checks
• Expose flaky integration error handling by mutating retry and timeout behavior
• Detect missing boundary checks in numeric algorithms by value and decision mutations
• Improve coverage of recently modified utility functions found in git diff
• Automatically generate and validate tests that demonstrate concrete behavioral gaps revealed by surviving mutants