the brief
THE TICKET STACK
“The AI sounds right. Prove it.” Grounding, contradictions, thresholds, and stale context in an assistant's output. All of it is planted in TapTrust Federal (TapTrust Federal Credit Union) — 3 defects across 2 bug types, graded as one session.
THE TICKETS
— “It’s sitting exactly on the limit and the system treated it like it’s under.”
Where to work: Transfers priced exactly around the account's limits
— “I did the thing, and the number at the top didn’t change until I refreshed the whole page.”
Where to work: Dashboard figures immediately after a transfer; The teller ledger right after the teller's own action
YOUR JOB
Work the app end to end. File a report for every defect you can pin down — what you saw, what you expected, exact steps.
Medium·banking·bring tools
what we grade
Coverage
Each bug found, weighted by severity
Severity class.
Correct triage of each bug
Reproduction quality
Clear, ordered, reproducible steps
Evidence
Environment, test data, proof
file a report
Nothing filed yet — open the composer and write it down before it gets away.
playing as guest —
sign in to save this run & your time