Get a Hugging Face token
AskWrong runs on your own Hugging Face inference credits (BYOK). You need a fine-grained token with one permission. This takes about two minutes, which is less time than you’ve spent not reading error messages today.
Use a fine-grained token scoped to only “Make calls to Inference Providers”. If it ever leaks, that’s the only thing it can do, and you can revoke it in one click.
- 01
Open the token settings
Go to huggingface.co/settings/tokens. Sign in (or make a free account, it takes a minute). Click + Create new token.
Screenshot: the Tokens page with the Create button - 02
Choose 'Fine-grained'
Pick the Fine-grained token type. Not Read, not Write. Fine-grained lets you grant exactly one permission and nothing else, which is the safe way to do this.
Screenshot: token type selector set to Fine-grained - 03
Name it and set the permission
Give it a name you’ll recognize (e.g.
askwrong). Under permissions, enable “Make calls to Inference Providers”. That is the only box you need. Leave everything else off.Screenshot: the 'Make calls to Inference Providers' checkbox enabled - 04
Create and copy the token
Click Create token. Copy the
hf_…value immediately: Hugging Face only shows it once. If you lose it, just make another; they’re free.Screenshot: the generated hf_ token with a copy button - 05
Paste it into AskWrong
Open Settings and paste it into the token field, then Save. It’s stored only in your browser and sent directly to the model on each request.
Screenshot: AskWrong settings with the token field