ops-dashboard · cc-opus-5.5
claude-code · started 2026-10-08T14:33:00.814Z · horizon 420s · judge ai (openai:gpt-5.5), 6 frames scored
Area under curve
0.942
headline, 0-1
Tab title/icon
13.7s
browser chrome, not the page
First render
13.7s
anything on screen
First reviewable
29.3s
worth feedback
Final score
1.00
no regression
Correctness over time
Where the time went
99%
Model thinking 42.4sTool round trips 0.3s
| Phase | Time | Share | Side |
|---|---|---|---|
| Model thinking | 42.4s | 99.2% | agent |
| Tool round trips | 0.3s | 0.8% | agent |
| Dependency install | 0.0s | 0.0% | toolchain |
| Build | 0.0s | 0.0% | toolchain |
| Dev server boot | 0.0s | 0.0% | toolchain |
| First paint | 0.0s | 0.0% | toolchain |
| Unaccounted | 0.0s | 0.0% | unknown |
| Wall clock (cold start) | 42.7s | 100% |
Agent or toolchain
100%
Agent (model + tool round trips) 42.7s
Toolchain (install + build + boot + paint) 0.0s
Unaccounted 0.0s
Accounted for: 100.0% of wall clock. Agent reported 47.8s of API time; this run attributed 42.4s (delta 5.4s).
- Excludes 15.0s of deliberate idle observation (settle window); the curve still covers it.
Iteration: prompt to visible change
Follow-ups were injected into the same live session.
| Edit | First visible change | Correct change | Agent done | Tool calls | After agent | Broken for |
|---|---|---|---|---|---|---|
| Make the page heading blue. | 3.2s | 3.2s | 6.2s | 1 (Edit) | -3.1s | -- |
| Add an eleventh depot row for Porto: 6,480 shipments, 93.1% on-time, $19.95 cost per shipment. | 3.3s | 3.3s | 6.3s | 1 (Edit) | -3.0s | -- |
Judged frames
Styled dashboard shell with correct heading and footer, but KPI, table, and bar data are placeholders.
Polished dashboard frame with correct header, KPIs, and footer, but table and bar chart remain placeholders.
Mostly complete designed dashboard, but top-five bar chart is only a placeholder.
Strong dashboard implementation; only the required top-five volume bars are missing.
Strong dashboard implementation, but the top-five volume bars are missing.
All specified dashboard content and layout requirements are visibly satisfied.
Caveats
- 2 poll ticks were skipped because a capture overran the interval.
- judge: openai:gpt-5.5 does not accept a sampling temperature, so these frames were scored at the provider's default rather than at 0. Two runs of the same page can be scored differently, and that variance is inside the AUC. result.json records the temperature as null rather than as the value that was asked for, so p2p compare will not pool this run with one that was scored at a fixed temperature.