AgentDrift
A single-file dashboard that flags when an AI agent's final synthesis contradicts the evidence its own tool calls gathered.
AgentDrift detects reasoning drift — when a multi-step agent gathers consistent evidence across its tool calls, then writes an optimistic summary that ignores what those tools actually found. It scores each run on a 0–100 Drift Score and surfaces HIGH/MEDIUM contradictions in a clean diff panel that pairs every claim against the contradicting step. Pure HTML with zero dependencies and zero build step; ships with a demo scenario plus a JSON log format so you can audit your own agent's reasoning chain.
Build log
Get an email when I ship a new prototype or essay. No funnel — just the work.