Usage instructions

[dotemacs]

Hello

I could be exposing my ignorance here on how Hugging Face and its API is meant to be used...

Registered with Hugging Face, got their Auth token.

Tried issuing some curl based commands like:

curl https://router.huggingface.co/hf-inference/models/Qwen/Qwen3-235B-A22B/v1/chat/completions \
     -H "Authorization: Bearer ${HF}" \
     -H 'Content-Type: application/json' \
     -d '{
        "messages": [
            {
                "role": "user",
                "content": "What is the capital of France?"
            }
        ],
        "max_tokens": 512,
        "model": "Qwen/Qwen3-235B-A22B",
        "stream": false
    }'

and that works. As in, I can authenticate correctly.

I've tried this:

(use-package starhugger
  :init
  (setq starhugger-hugging-face-api-token (auth-source-pick-first-password :host "huggingface.co"))
  :config
  (setq starhugger-auto-idle-time 0.05
       starhugger-completion-backend-function #'starhugger-hugging-face-inference-api
       starhugger-model-id "codellama"))

I open a sample buffer foo.py and enter print followed by M-x starhugger-trigger-suggestion only to see the following in *Messages* buffer:

You can run the command ‘starhugger-trigger-suggestion’ with M-x st-t-
error in process filter: could not parse JSON stream: 2, 1, 2 [2 times]

What am I doing wrong?

Thanks

[daanturo]

Hi, thank you for using my package.

https://router.huggingface.co/hf-inference/models/Qwen/Qwen3-235B-A22B/v1/chat/completions

Looks like HF has updated their text inference API to be OpenAI-compatible and the old one supported by starhugger is no longer available.

I'm planning to add support for OpenAI-compat APIs, it should allow using a bunch of other providers, too.