todo-app-scaffolded · pi-sonnet-5.5
pi · started 2026-10-07T15:22:39.687Z · horizon 480s · judge ai (anthropic:claude-sonnet-5), 1 frames scored
Area under curve
0.940
headline, 0-1
First render
28.9s
anything on screen
First reviewable
28.9s
worth feedback
Final score
1.00
no regression
Correctness over time
Where the time went
97%
Model thinking 27.6sTool round trips 0.0sUnaccounted 0.7s
| Phase | Time | Share | Side |
|---|---|---|---|
| Model thinking | 27.6s | 97.5% | agent |
| Tool round trips | 0.0s | 0.0% | 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 | 0.7s | 2.5% | unknown |
| Wall clock (cold start) | 28.4s | 100% |
Agent or toolchain
97%
Agent (model + tool round trips) 27.7s
Toolchain (install + build + boot + paint) 0.0s
Unaccounted 0.7s
Accounted for: 97.5% of wall clock.
- 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 header background blue. | 4.2s | 4.2s | 9.7s | 0 | -5.5s | -- |
| Add a fourth column called "Blocked" after Done. | 3.1s | 3.1s | 5.3s | 2 (edit) | -2.2s | -- |
Judged frames
All criteria met: polished board with columns, cards, badges, and styled header/button.
Caveats
- The page had only 4 distinct visual state(s) across 74 frames, so frames/ holds 4 screenshot(s) and the other 70 frames reuse them. That is deduplication, not missing captures: the timeline in result.json maps every frame to its picture.
- judge: anthropic:claude-sonnet-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.