run-suitelisted

# Skill: run-suite Run the right tests, against the right target, with output you can act on. Two failures are common and both waste real time: running the wrong target and reporting a false green — a suite that "passed" because a filter matched nothing exits zero and looks identical to success. ## Preconditions `.claude/test-profile.md` exists and lists the run commands. If it does not, run **probe-conventions** — guessing a target name usually produces "no tests found", which reads as success. ## Procedure ### 1. Resolve what to run | The user said | Run | | --------------------------------- | --------------------------------------------------------------- | | "run this test" after an edit | that one file, no filter | | "run the API tests" | that group's directory | | "run the suite" / "run everything" | everything, at the repo's default target | | something ambiguous | ask — but ask once, with your recommendation, not a form to fill | When the context makes it obvious — they just edited one spec — run that and say what you chose. Do not interrogate someone about a device matrix when they asked you to check their last edit. ### 2. Resolve where to run it Targets (Playwright projects) are named per the profile. If the profile says targets are generated r