Thanks for an amazing project! Tested with VSC today and worked amazingly well at agent mode for tool calls! Fantastic!
I had some issues in continue.dev previously on pycharm.
It would be great for (a) curiousity (b) disabling tool-less models if there was:
detailed - chronological - logging?
Can we make it so everytime a request is made it logs which model it was and maybe how many tokens?
Just when you ask "which underlying model are you?" VSC just said it was github copilot
My logs don't really help:
[server] [Proxy] This operation was aborted from DeepSeek V4 Pro (NV), falling back (attempt 1/20)
[server] [Health] Check complete.
[server] [Proxy] NVIDIA NIM API error 503: Service Unavailable from GLM-5.1 (NV, slow cold-start), falling back (attempt 1/20)
[server] [Proxy] OpenRouter API error 404: This model is unavailable for free from Kimi K2.6 (OR free), falling back (attempt 2/20)
[server] [Proxy] Google API error 400: Invalid JSON payload received. Unknown from Gemini 3 Flash Preview, falling back (attempt 3/20)
[server] [Health] Checking 6 keys...
[server] [Health] Check complete.
[server] [Health] Checking 6 keys...
[server] [Health] Check complete.
or
google | Google API error 400: Invalid JSON payload received. Unknown name "enumDescriptions" at 'tools[0].function_declarations[42].parameters.properties[3].value': Cannot find field. | 21:22
openrouter | OpenRouter API error 404: This model is unavailable for free. The paid version is available now - use this slug instead: moonshotai/kimi-k2.6 | 21:22
nvidia | NVIDIA NIM API error 503: Service Unavailable | 21:22
nvidia | This operation was aborted | 21:17
google | Google API error 429: Your prepayment credits are depleted. Please go to AI Studio at [redacted-url] to manage your project and billing.
There is a record of which worked etc but not in chronlogical order for user debugging.
Thank you!
James
Thanks for an amazing project! Tested with VSC today and worked amazingly well at agent mode for tool calls! Fantastic!
I had some issues in continue.dev previously on pycharm.
It would be great for (a) curiousity (b) disabling tool-less models if there was:
detailed - chronological - logging?
Can we make it so everytime a request is made it logs which model it was and maybe how many tokens?
Just when you ask "which underlying model are you?" VSC just said it was github copilot
My logs don't really help:
[server] [Proxy] This operation was aborted from DeepSeek V4 Pro (NV), falling back (attempt 1/20)
[server] [Health] Check complete.
[server] [Proxy] NVIDIA NIM API error 503: Service Unavailable from GLM-5.1 (NV, slow cold-start), falling back (attempt 1/20)
[server] [Proxy] OpenRouter API error 404: This model is unavailable for free from Kimi K2.6 (OR free), falling back (attempt 2/20)
[server] [Proxy] Google API error 400: Invalid JSON payload received. Unknown from Gemini 3 Flash Preview, falling back (attempt 3/20)
[server] [Health] Checking 6 keys...
[server] [Health] Check complete.
[server] [Health] Checking 6 keys...
[server] [Health] Check complete.
or
google | Google API error 400: Invalid JSON payload received. Unknown name "enumDescriptions" at 'tools[0].function_declarations[42].parameters.properties[3].value': Cannot find field. | 21:22
openrouter | OpenRouter API error 404: This model is unavailable for free. The paid version is available now - use this slug instead: moonshotai/kimi-k2.6 | 21:22
nvidia | NVIDIA NIM API error 503: Service Unavailable | 21:22
nvidia | This operation was aborted | 21:17
google | Google API error 429: Your prepayment credits are depleted. Please go to AI Studio at [redacted-url] to manage your project and billing.
There is a record of which worked etc but not in chronlogical order for user debugging.
Thank you!
James