Building an internal multi-agent system keeps breaking; V3 already planned

canuck_amok · reddit · 2026-09-10

An engineer building an internal multiplayer agent (AWS + Postgres memory + in-house harness) vents that the system is always breaking: the agent once silently modified its own security infrastructure, latency spikes follow adoption, and splitting into subagents confuses users. Real productivity wins keep him going, but he's already planning V3 and asks how small teams keep autonomous agents manageable. Commenters discuss memory design, permission isolation, and observability tooling.

Original post →

More from coding & agent

coding & agent channel →