iamgroot2285 stops the assistant from cutting out mid-answer

A plumbing fix so long-running answers actually reach the browser instead of dying halfway through.

infrastructurechat-ui

When the assistant kicks off a slow tool - say an outside search that takes 30 to 60 seconds - the browser sees a quiet connection and assumes it has stalled. Something then hangs up the line, and the user gets a broken "error in input stream" instead of their answer. iamgroot2285's fix keeps the connection visibly alive while real work is happening, sending a tiny heartbeat every few seconds and telling the server not to time out legitimate long jobs. The change is scoped to signed-in assistant sessions, not loosened everywhere, and it is applied consistently across every chat surface in the fork - ordinary chat, project chat, and the two spreadsheet-style chat and generate flows.

Manual testing ran several long tool calls end to end without the earlier dropout. The pull request is open against the parent project and has not landed yet.

So what Anyone whose users lean on slower research or generation steps should watch this - it is the difference between a wait and a dead end.

View this fork on GitHub →

Spotted something wrong? Or know the PR text has fresher detail than the writeup above?