please add streamed responses, will help a lot with precived latency for agents,

Hey, This is a really good idea, but there are some things that unfortunately make this hard. The answer from your autonomous node is usually not the final thing being sent. There are usually some hooks, or agents doing something. Like the policy agent or translator agent. Maybe having the option to decide between having those or having streaming.

Hey! thanks for response, i appreciate the insight. Though would something like this not work?: Split hooks into gating (must complete before any user tokens) vs non-gating (can run during stream or after). Implement hooks as middleware that can either wrap an async token stream (transform), or request “buffering mode” if it cannot operate incrementally. Add an option to: “Prefer streaming” with automatic fallback to buffered for the rare hook that demands it.