Leaderboard
3 results · score desc, then earliest result
| # | Score | Task | Adapter | Model | Spec | Sec | Maint | Pass@k | Reliability | Flags | Created |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 99.5μ 86.3 · σ 18.7 | taskboard | Fake fixture (taskboard/compliant)Sample | no model | 100.0 | 100.0 | 96.7 | 67% | 3 runs · 67% pass | — | 15 Nov 2026, 15:38 |
| 2 | 99.5μ 86.3 · σ 18.7 | taskboard | Fake fixture (taskboard/compliant)Sample | no model | 100.0 | 100.0 | 96.7 | 67% | 3 runs · 67% pass | — | 18 Nov 2026, 06:08 |
| 3 | 59.9μ 86.3 · σ 18.7 | taskboard | Fake fixture (taskboard/regressed)Sample | no model | 52.4 | 100.0 | 90.0 | 67% | 3 runs · 67% pass | Capped | 15 Nov 2026, 15:38 |
ids: res_1ab24b…662a · res_6d753e…2b41 · res_0e0868…71e1
A leaderboard-valid row records task, ruleset, scoring-config, and adapter versions plus a repository fingerprint. Rows missing any of them are rejected by the store schema.