Skip to content
m4bwav edited this page Nov 21, 2026 · 1 revision

Does it need anything installed?

Python 3.9 or later and the project's own test runner. diff-cover is used when present.

Why did run say "no runnable suite detected"?

detect found no dotnet project, no Vitest or Jest dependency and no pytest marker, or found only Unity or Luau, which run does not start. A folder of plain test_*.py files with no config is detected as of 0.1.0. See Stacks and runners.

My repo has two test stacks. Why did only one run?

run runs the first runnable stack it detects. Run the other yourself and pass both reports to parse.

changed says every line is covered, but I know it is not.

The coverage paths probably do not match the repository's paths, so no changed line was found in the coverage file. Fix the paths in the coverage report and run again.

Is a flaky test a pass?

No. run --reruns names it, and the skill reports it as a bug to fix.

Will the skill delete a failing assertion to get green?

No. Its rules forbid weakening or deleting a failing assertion, and skipping or quarantining a test without saying so.

Does lint replace mutation testing?

No. It catches assertions that cannot fail from their text. Mutation testing (Stryker.NET, StrykerJS, mutmut, hand mutants for Unity and Luau) shows whether the tests catch real breakage.

My stack is not one of the five. Does it still help?

Yes. The risk-first plan, lint's patterns and parse (any runner that writes JUnit XML) work beyond the five. run and detect start only the runners listed in Stacks and runners in 0.1.0; for anything else, run the suite yourself and give its report to parse. What the agent learns about a new stack goes into the plugin's LEARNINGS and references.

Browser end-to-end tests?

They are in scope as testing. In 0.1.0 the plugin has no browser notes of its own and its skill description sends browser runs to Playwright; skill evals go to evergreen-test.