Skip to content

Stacks And Runners

m4bwav edited this page Nov 21, 2026 · 1 revision

testwright covers any stack; this page lists the ones tw.py automates in 0.1.0. For any other stack, the general method still applies: run the suite yourself and pass a JUnit XML, TRX, NUnit or Jest/Vitest JSON report to parse. New stacks are added to the references as the plugin meets them.

tw.py detect looks for each stack in this order and can report several:

Stack Found by Command run uses
unity ProjectSettings/ProjectVersion.txt none: run never starts Unity
dotnet a .csproj, .sln or .slnx file, and no Unity project dotnet test --logger trx --results-directory TestResults
js vitest, else jest, in package.json dependencies or devDependencies npx vitest run --reporter=json --outputFile=tw-results.json or npx jest --json --outputFile=tw-results.json
python pytest.ini or conftest.py, pytest in pyproject.toml, or any test_*.py or *_test.py file python -m pytest --junitxml=tw-results.xml
luau default.project.json, wally.toml or any .luau file none

run picks the first stack in that list that has a command, so a repository with both a .NET project and Python tests runs only dotnet test. Run the other suite yourself and pass its report to parse. Report paths are rewritten to a temporary folder.

Unity and Luau

run starts neither. The skill points the agent to the stack's reference for the headless command, then runs tw.py parse on the results file:

  • unity.md: Unity Test Framework in batch mode; results are NUnit 3 XML, which parse reads.
  • luau.md: jest-lua or TestEZ.

The other references are dotnet, js, python, quality (mutation testing), legacy and games.

Only the Python path was run for this wiki; dotnet, Vitest, Jest, Unity and Luau runs are not tested here.

Clone this wiki locally