Every failure is classified — product defect, agent shortcoming, environment problem or assertion mismatch — so a confused agent never arrives as a bug report.
An autonomous tester that reports every failure as a defect is worse than no tester, because someone has to triage the noise. Argus classifies each failure by cause and has the agent declare why it gave up, so a suspected product bug is flagged the moment it happens and an agent that simply got lost is recorded as exactly that.
Failures are classified by cause, not just recorded.
Product defect, agent failure, environment failure and assertion failure are distinguished.
Loop, step-limit and timeout protections are separate outcomes.
The agent declares why it stopped.
An agent getting lost is never filed as a software defect.
Classification plus a rule: the AI never asserts a defect on its own authority. A human reviews and files it.
A ready-to-file report — environment, build stamp, the agent’s diagnosis, assertion results and step-by-step repro.
Yes, environment failures are their own class, and a failed preflight aborts before any model call.
A scenario is an objective, a role and assertions — no selectors, nothing to rewrite after a redesign.
Creator, approver and supplier each get a fresh, independently authenticated session.
A passing run distils its path into intents, never coordinates, for later runs to reuse.
Every model call is metered and aggregated from call to suite.
See the complete product: problem, features, how it works and deployment.
Request a demo to watch a multi-role regression suite run through your real interface.