promptdrift it-suite  ·  1 regressed · 1 unstable · 1 new · 1 removed · 1 stable  ·  score 89% → 58%
model mock                                                                                                    
┏━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━┳━━━━━━━┳━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ case     ┃ assertion            ┃ before ┃ after ┃ verdict      ┃ note                                     ┃
┡━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━╇━━━━━━━╇━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ faq      │ contains-welcome     │      — │   3/3 │ 🆕 new       │ no baseline recorded for this assertion  │
│ greeting │ contains-hello       │    3/3 │   0/3 │ 🔴 regressed │ all 3 samples failed (baseline 3/3)      │
│ legacy   │ (removed from suite) │        │       │ ➖ removed   │                                          │
│ refund   │ contains-申请退款    │    3/3 │   1/3 │ ⚠️ unstable  │ samples disagree (1/3 passed) where      │
│          │                      │        │       │              │ baseline was 100%                        │
│ tone     │ regex-polite         │    2/3 │   3/3 │ ✅ stable    │                                          │
└──────────┴──────────────────────┴────────┴───────┴──────────────┴──────────────────────────────────────────┘
❌ gate FAILED — 1 regressed case(s)

