Backend timeout errors with local backend

[SynthZero]

What happened?

The plugin spams backend warnings while the model still works.

Describe the bug

The backend timeout is set too low in the plugin. When running a loccal backend, it constantly shows timeout warnings while the model still works. If the backend was just started, time to first token is always longer than usual.
Can you expose the option as a VScode setting, or something?

Steps to reproduce

1. run a local backend, like llama.cpp
2. make the first request

Expected behavior

I expected there to be a manual setting like in Copilot Chat.

Environment

• VS code 1.120.0
• OS Ubuntu 24.04
• AgenticFlowX 2.0.15

[rixrix]

@SynthZero thanks for reporting this. I've got the fix pushed in latest release.

I've tested on Llama.cpp with Gemma.

you can set the timeout here

• Settings -> Behaviour -> Model warm-up timeout. default to 60s

can you please confirm that it's working for you once you get the chance. thank you