Demo 03 · Simulated

Agentic trace viewer

Debugging an agent system is mostly reading traces like these — knowing which step burned the tokens and which retrieval went wrong before touching the code. Pick a scenario and replay the agent loop step by step: the tool choice and the reasoning behind it, the retrieval hits behind each observation, and the step budget counting down.

Every retrieval runs over the same 12-passage corpus as the search demo, passages about my own projects — read the case studies →

These traces are hand-authored, not recordings of a live model run. They illustrate how the agent loop behaves — tool choice, retrieval, budget pressure, recovery — and the data shape matches a real recorded run, so recordings can replace these files later. Latency and token counts are authored estimates.