Stops AI agent run timelines from accidentally leaking secret keys during context resets.
Built by Parcha Labs (GREP.AI, YC F26) and newly open-sourced, AgentRun has 959 GitHub stars and is used by developers running AI coding and automation agents.
When an agent's memory resets or summarizes, secret keys it holds are now scrubbed from the activity timeline, preventing accidental credential leaks in logs and live views.
WHERE In the agent run timeline view, where context resets and summarized older context appear as phase rows.
SCENE You are watching an agent run's timeline. Halfway through, it resets its context and summarizes earlier steps. If a held key was mentioned, will the timeline scrub it or show it?