🔑 An API key buried in `.env` can tell an agent what it's allowed to do, but it can never tell you who actually asked. When every user shares the same bot account, every GitHub issue and every internal ticket gets logged under one anonymous identity, with no trace of the real person behind it. Rotating a key used to mean editing code and redeploying. Wanting OAuth meant building your own authorization URLs, token storage, refresh logic, and consent screens from scratch. That unglamorous but heavy plumbing is exactly what LangChain's new Connections feature for Managed Deep Agents takes off your plate. Credentials can be agent-owned (shared by everyone) or user-owned (resolved per caller), and both static API keys and OAuth grants are supported. A shared key covers something generic like web search, while a GitHub integration can resolve to each user's own permissions and identity — so the issue it creates is logged under their real name, not "bot." Connections: Managed credentials and per-caller identity for Managed Deep Agents Separating "what credentials exist" from "who's calling" feels like it could quietly kill a lot of tedious OAuth plumbing in agent development. #AIAgents# #LangChain#