microsoft
Enable Claude 3.5 Sonnet (Preview) for all clients
Type: Bug
Enable Claude 3.5 Sonnet (Preview) for all clients
It's asking me this for every chat, it loose context of the files after that ENABLE button
Extension version: 0.22.1 VS Code version: Code 1.95.1 (65edc4939843c90c34d61f4ce11704f09d3e5cb6, 2024-10-31T05:14:54.222Z) OS version: Windows_NT x64 10.0.22621
Oh wow, I kinda don't like copilot to publish the issue with all my private content as a title. That was unexpected.
Hmm that's not expected. Are you saying even after restarting VS Code you see it ask you to enable the policy every time?
The policy prompt kept appearing repeatedly, despite the ENABLE state confirmed in my Copilot settings here: https://github.com/settings/copilot.
After restarting VS Code, Claude now seems unresponsive, stalling on "Generating....", so I can’t verify if the policy issue is still there. However, I suspect it may be resolved, as the policy acceptance message did appear before any interaction with Claude.
That said, a full VS Code restart shouldn’t be required after clicking "ENABLE" just once.
That said, a full VS Code restart shouldn’t be required after clicking "ENABLE" just once.
You are correct it should not, I will see if I can reproduce it but haven't been able to yet on my end. I was trying to offer a solution to the issue you were experiencing.
Did you enable the policy initially from within VS code or from the website?
Sorry about the tone of that sentence, it's seems chatGPT is intentionally rewriting my answer in a less friendly tone than my messed up English.
I enabled it from within VS Code. Timeline:
click on Copilot Edits > select model > see that there is now Claude. > clicking on Claude > writing my message, adding context > send
Policy message from vs code > Click enable > Answer from Claude ( but it's unaware of any context, as if the policy message removed any contextual information/linked files )
So I clicked on the + to create a new chat, and the whole exact same thing happen.
Then I've restarted, and tried Claude again, but I had the "Generating..." stale issue.
And as of now, instead of a stale :
But I suppose it's more of a bandwidth/api usage issue since a lot of people are trying the model at the same time.
If you paste that request Id here I can look into what is happening. Thank you for the bug report. I will keep it open and try to reproduce.
I experienced that issue too just now, "Enable Claude 3.5 Sonnet (Preview) for all clients" in every chat under gpt-5mini, but seems to disappeared when switch to claude 3.5 manually
