Live Prompt Archive / Multi-Model
Inspect every prompt, counterexample, and response across the live runs.
Select a model and inspect the exact repair loop: prompt, verifier counterexample, raw completion, and the final accepted code when convergence happened.
Explorer
Tool-by-tool trajectories
Pick a model and tool to see where the loop corrected itself, where it repeated a mistake, and where the budget ran out.