The HN post with 689 points about an agent that destroyed a production DB is generating massive searches. I'm not rehashing someone else's war story — I'm opening my own CrabTrap and async agent logs to show what destructive operations almost made it through, where the trust chain broke, and why fra
The HN post has 127 points and nobody's talking about the real problem: when an async agent fails, you don't get a stack trace. You get silence. And silence in production is the worst bug that exists.
I installed CrabTrap on my real infrastructure — a proxy that intercepts HTTP calls from agents and judges every response with another LLM before executing it. I measured latency, false positives, and token cost overhead. The result has a circular trust problem that nobody in the announcement mentio
32 years in the dev trenches. Here I write what I learned, what I broke, and what nobody tells you in the tutorials.
No spam. Unsubscribe anytime.