test-changes

To get a focused error report, this skill dispatches a dedicated test runner agent to execute your build, test, and lint commands, filtering the output to display only errors and concise summaries while keeping full logs external.

Yes, you can customize test commands by creating a .claude/test-config.yml file in your project root, which the skill reads to execute specific build, test, and lint instructions instead of relying on auto-detection.

Running tests and lint with a sonnet-based agent keeps context clean by executing the suite in a separate runner and returning only a filtered error report with context lines and return codes to the main session.

The summary includes concise results for your build, tests, and lint, featuring filtered error outputs with relevant context lines and return codes, while specifying where to find the full external logs.

Yes, the test runner supports both standalone execution and run-phase workflow integration, including optional state handling to fit seamlessly into your existing automated build and test processes.

You need a separate agent to run tests and lint because executing these commands directly often produces excessive noise, whereas a dedicated runner outputs a filtered error report to maintain a clean main context.