Quickstart

Once installed, drive everything from seven slash commands:

/qa-my-app:init                 # First-time bootstrap; builds QA-tests/
/qa-my-app:status               # Read-only health + inventory snapshot
/qa-my-app:sync                 # After code changes, reconcile catalog
/qa-my-app:scan                 # Force full rescan (backs up tasks first)
/qa-my-app:run T03              # Execute one task end-to-end
/qa-my-app:run-all              # Execute every task — parallel runners
/qa-my-app:run-all T01,T03,T07  # Subset by task id
/qa-my-app:run-all /customers   # Subset by route prefix
/qa-my-app:run-all failed       # Re-run only tasks whose last result was FAIL/BLOCKED
/qa-my-app:run-all changed      # Re-run only tasks whose route source is dirty
/qa-my-app:verify               # Test just what changed (conversation + uncommitted diff)
/qa-my-app:verify PROJ-123      # Verify against a ticket's acceptance criteria
/qa-my-app:verify --branch      # Verify the whole PR (everything different from main)

You don’t always have to type them

status and verify are model-invocable — ask in plain language and Claude reaches for them:

"is QA set up in this repo?"      → /qa-my-app:status
"verify what I just changed"      → /qa-my-app:verify
"does my change actually work?"   → /qa-my-app:verify
"check PROJ-123 is done"          → /qa-my-app:verify PROJ-123

The other five stay strictly manual. init, sync, and scan rewrite your catalog, and run / run-all spawn real browser runs — Claude will never start any of those on its own.

The live run dashboard

Every /qa-my-app:run, /qa-my-app:run-all, or /qa-my-app:verify invocation writes a self-contained dashboard at QA-tests/results/runs/<runId>/report.html. Open it in a browser while the run is in flight — the page meta-refreshes every 3 seconds and shows the queue draining live (pending → dispatched → complete, with per-task verdicts, defects, and links to result.md). Once the run finishes, auto-refresh disables itself and the same file becomes the canonical browse view for that run.

No server, no install, no dependencies — double-click to open.

Typical first session

/qa-my-app:init        # discover routes, analyze pages, author tasks (5–20 min first time)
/qa-my-app:run-all     # execute the whole catalog in parallel

Subsequent runs are incremental — /qa-my-app:sync only re-analyses routes whose source fingerprint changed. See How it works for why this stays deterministic and PR-reviewable.