RockB — AI tools and engineering guides

ToolReplay: Auditing an Agent Session for Non-Determinism and Scope Overreach

If you let a coding agent run unattended, the only durable record of what it did is the transcript of its tool calls. Dashboards show you totals; the transcript shows you the sequence. ToolReplay is a small, dependency-free Python CLI that reads one such transcript and answers three narrow questions about it: did the same call ever return two different answers, did the agent repeat a call that could not have changed anything, and did it call a tool it was never granted. It reads a JSON Lines file, and it does the work offline. ...

October 8, 2026 · 12 min · baeseokjae
Density AI Agent Flight Recorder

Density AI Agent Flight Recorder: Storage-Optimized Observability for Agent Traces

Density is an open-source flight recorder for AI agents that solves the growing problem of agent trace storage bloat. Instead of deleting traces to save space, Density compresses them into a tiered storage system — HOT, WARM, and COLD — with verifiable recall guarantees at each tier. It achieves up to 50.6x compression on traces and 32x on embeddings while maintaining recall above 0.99, making it the first storage-layer observability tool that lets you keep everything without guessing what you will lose. ...

August 6, 2026 · 12 min · baeseokjae