hv-qalisted

QA the built product — not the diff. Use on "/hv-qa", "run QA", "test the feature", "validate the build", before ship as a gate, or on first cycle to scaffold a per-repo strategy. Detects testing surfaces per repo (web, API, CLI, mobile, lib), picks runners (Playwright, smoke, contract, lighthouse, ZAP, axe), and produces a scored report with executable pass/fail results plus audit-style usability findings. Strategy is per-repo in .hv/qa/<repo>.md so the skill never hardcodes "browser". Modes — first-run (probe + propose strategy), run (execute strategy, emit verdict), restructure (audit strategy files). Opt-in gate via ship.qa.
l4ci/hv-skills · ★ 0 · Testing & QA · score 73

Install: claude install-skill l4ci/hv-skills

**Print the banner below verbatim before any other action — skip if dispatched as a subagent.** See `references/banner-preamble.md`. ``` ════════════════════════════════════════════════════════════════════════ 🧪 hv-qa · QA the built product (not the diff) triggers: "qa this", "kick the tires" · pairs: hv-review, hv-ship ════════════════════════════════════════════════════════════════════════ ``` # hv-qa — Product Quality Assurance `/hv-qa` and `/hv-review` are deliberately separate: - **`/hv-review`** answers *"does this diff make sense"* — reads commits + diff, no execution. - **`/hv-qa`** answers *"does the product work"* — runs tests, probes, scans against the built artifact. They never call each other. `/hv-ship` may call both, each behind its own config flag. ## Configuration Read `.hv/config.json`: - `models.orchestrator` — model dispatching the runners (default `opus`). - `qa.gate` — `"advisory"` (default) emits verdict, never blocks ship; `"blocking"` causes `ship.qa: true` invocations to halt on `FAIL`. - `qa.afterWork` — `false` (default). When `true`, `/hv-work` invokes `/hv-qa run` post-cycle if touched files match a QA target's `Watch globs`. - `ship.qa` — `false` (default). When `true`, `/hv-ship` calls `/hv-qa run` between `/hv-review` and the merge/PR step. ## When to Use - *"QA this"*, *"kick the tires"*, *"does this actually work?"* — manual exploratory run. - After a feature lands and before opening a PR, when you want product-level evid