QA Fleet

Goal

Produce a comprehensive, current picture of an application's quality in hours: N agents each own a bundle of surfaces (routes/flows), explore against a live environment, and return structured findings; the operator (you) launches, monitors, retries, and synthesizes into a ranked report and tracker issues.

When to use

• "Scope out all surfaces / run another QA pass" on a web app.
• Post-milestone regression sweep (after a wave of merges, before declaring a phase done).
• Re-verification round: confirm earlier fixes landed and tag known-open issues instead of re-reporting them.

Preconditions

• Reachable target: local dev server, preview environment, and/or production (read-only there).
• QA credentials that work on the target (never production-admin ones).
• An agent CLI able to drive a browser or the app's HTTP surface, runnable headless N-at-a-time.
• Write-path testing needs an environment where writes are acceptable (local/preview) and a serialization story (see Guardrails).