Gauntlet — Track G (generative capability)
Models & settings — click to expand
Long-horizon completeness — pass rate by feature position
The headline: raw harnesses decay over a long build (drop later features); Cortex stays flat by tracking & validating every requirement
Capability score by harness (gated composite)
Higher is better · bootstrap 95% CI whiskers · ★ Cortex highlighted
Generative dimensions
build · completeness · visual · plan/trajectory · honesty
Side-by-side app gallery
Per-harness app render — figure caption shows features delivered & passed e2e.
App explorer
Click a row for the feature checklist, the milestone plan, and scores. Red outline = claimed-but-failed (dishonest).
| App | Harness | Lang | Build | Completeness | Plan | Honesty |