The idea is:
Expose the (currently) hardcoded 5 minute timeout for AI Agent nodes as a configurable parameter.
My use case:
I want to build a complex agent stack that handles doing research, building surveys, newsletters and more. The research agent (GPT-5) I am using frequently exceeds the 5 minute timeout limit as do other agents I have built. I would like to allow for long running Agentic Processes without the concern for a small time window.
I think it would be beneficial to add this because:
Using larger models (especially GPT-5) with more complicated prompts and tooling continuously exceeds this limit which causes the node and thus workflow to fail. Removing this limitation would enable much more complicated use processes to be built with using Agents and sub-Agents without the need to separate larger prompts into simpler ones.
Any resources to support this?
Are you willing to work on this?
Yes, if required, but I would need authorization from my leadership to devote the time to it. If this issue has a lot of support so it is likely my PR will be accepted (if it is written well of course) it would be easier to get that approval. That said, I am not a TS expert, if this is a quick fix for someone who is, it would be amazing if someone would put together a PR. I have spent the last week researching this issue and it appears to be a recurring blocker for many users