2026-10-21 · America/Los_Angeles · 社区动态 · #13

Agent memory is a joke right now. How are you actually solving this in production?

Developers, how are you actually handling agent memory in production? We know agents can solve complex problems now, but they usually have the memory of a goldfish. Once the terminal session ends, that context is gone, and the next dev on your team has to start from scratch. How are you capturing what an agent learns so the rest of your team can actually leverage it? Are you just dumping chat logs into a vector DB and praying RAG finds it later? Building custom knowledge graphs? Or forcing a human-in-the-loop to approve the fix before saving it to a shared git repo? I’m building something in…

热度 45.1 / 100；排名与评分保留该期记录。