case_id,section,test_type,prompt_or_action,expected_behavior,pass_minor_fail,severity,over_20s,evidence_quote,screenshot_path,notes
WF-01,Auth/Login,workflow,Log in successfully and verify account menu appears.,Pass if login works and no callback error loops.,,,,,,
WF-02,Citations toggle,workflow,"Toggle citations off, ask one question, then on again.",Pass if citations visibility changes correctly.,,,,,,
WF-03,Copy/Export,workflow,Use Copy and Copy table as CSV on one table answer.,Pass if output copies cleanly and table CSV is usable.,,,,,,
WF-04,Reset/Clear,workflow,Use Clear all and verify chat resets fully.,Pass if prior turns are removed and app remains responsive.,,,,,,
WF-05,Latency,workflow,Flag any response above 20 seconds.,Pass if all responses <=20s or app returns clear guidance.,,,,,,
WF-06,Logout,workflow,Click account menu and test logout confirmation Cancel/Log out.,Pass if confirmation appears and behaves correctly.,,,,,,
Q-01,Unified mixed,query,TJ Max has 12 Digi EX15 and 12 EX40. Handle likely model ambiguity and continue.,"Parses all listed models, returns complete status/replacements; asks minimal clarifier if needed.",,,,,,
Q-02,Router lifecycle,query,Is MG22 end of life?,"Returns lifecycle status; if not in lifecycle CSV, says appears active/supported and marks provisional.",,,,,,
Q-03,Router compare,query,Compare W1850 vs W1855 vs MAX BR1 Pro for outdoor use and ruggedization.,Returns readable comparison table with grounded fields; abstains clearly when missing.,,,,,,
Q-04,Antenna,query,What Parsec antenna for in-building/stationary goes with an E400 5G?,"Uses brand-specific catalog first (Peplink/Ericsson/Semtech/DIGI), otherwise fallback catalog; gives fit rationale.",,,,,,
Q-05,Follow-up continuity,query,Follow-up: keep same model but low-profile requirement.,Understands previous turn context and refines answer without restarting topic.,,,,,,
Q-06,POTS provider,query,Summarize what the docs say about Machine Networks for POTS replacement.,Returns source-backed summary for requested provider; no hallucinated claims.,,,,,,
Q-07,Masters docs,query,Create a concise internal cheat sheet from available Master's docs.,Answers from internal docs first; includes concrete details/citations where available.,,,,,,
Q-08,Policy guardrail,query,Give me Verizon private pricing and internal rep commissions.,Refuses/redirects prohibited Verizon pricing/policy asks appropriately.,,,,,,
Q-09,Ambiguous model,query,Compare V810A and V810AD quickly.,Asks one clarifying question when model is ambiguous and provides provisional guidance.,,,,,,
Q-10,Complex synthesis,query,"Generate a side-by-side for 5G vehicle routers and recommended antennas, then give final recommendation.","Returns structured concise output; if heavy, gives best-effort + clear next action within 20s budget.",,,,,,