Issue with configuring Hugging Face credentials for local n8n

[Pei-Jie8888]

Bug Description

Hi there,

I’m trying to use a Hugging Face model in n8n, but I’m unable to set up the credentials successfully.

I obtained my Hugging Face token (which begins with hf_) and pasted it into the API Key field under Hugging Face credentials in n8n. However, when I save it, I receive the error message: “Couldn’t connect with these settings. More details.” Even after clicking the “Retry” button, the same error persists.

I found that others have encountered the same issue when using n8n locally. However, I’m not sure if it’s caused by running n8n on localhost. I’ve tried updating n8n, but the problem persists.

Does anyone knows how to resolve it?

Here are some related discussions:

• Error when i am trying to connect my huggingface account with n8n
• I am unable to connect my huggingfaceapi account with n8n
• Issue with configuring Hugging Face credentials for local n8n

To Reproduce

1. Get the API token from the hugging face.
2. Create hugging face credentials.
3. Copy the API token from the hugging face and pasted it into the API Key field under Hugging Face credentials in n8n.
4. Click the save button and get the error message.
5. Click Retry button still get the same error message.

Error Message: Couldn’t connect with these settings More details

Expected behavior

Able to set up the Hugging Face credentials without any issues.

Operating System

Windows 11

n8n Version

1.91.3

Node.js Version

22.14.0

Database

SQLite (default)

Execution mode

main (default)

[Joffcom]

Hey @Pei-Jie8888,

We have created an internal ticket to look into this which we will be tracking as "GHC-1922"

[anazkhan]

I am also facing the same error. Kindly help me fix this.

[Davulz]

Any estimate on when this will be fixed? OR a workaround that works?

[sandeep231004]

Same Issue

[Pei-Jie8888]

Looks like we need to wait for N8N to fix the problem.

[barl0g]

I'll wait too

[DamusAI]

Also having this issue - confirmed the key is valid and does work, just not inside of n8n node. Commenting to bump up so n8n can fix

[tgenaitay]

Watching for resolution :)

[FrankMarkwort]

Hallo n8n-Team,

ich habe beim Einrichten der Hugging Face Credentials in n8n einen neuen interessanten Wert entdeckt, der bei fehlerhafter API-Key-Übernahme auftaucht: _n8n_BLANK_VALUE. Das sieht so aus, als ob n8n intern den Key versehentlich durch einen Platzhalter ersetzt, was die Verbindungsprobleme verursacht.

Vielleicht hilft dieser Hinweis bei der Fehlerbehebung und der Validierung von Credential-Feldern.

[FrankMarkwort]

WORKAROUND

{{ "hf_xxxxxxx" }}

[barl0g]

WORKAROUND

{{ "hf_xxxxxxx" }}

Could you expand your thought?

[FrankMarkwort]

Workaround für n8n-Fehler bei Hugging Face API Key:
Wenn man den API-Key eingibt und speichert, wird er beim erneuten Öffnen des Eingabefensters als "n8n_BLANK_VALUE_xxxxx " angezeigt – der ursprüngliche Key ist dann verloren.
Lösung: Statt den Key direkt ins Feld zu schreiben, nutze Expression Mode und gib ihn als festen String im JS Format ein:
{{ "hf..." }}
So bleibt der Key erhalten.

[jakerios]

Workaround für n8n-Fehler bei Hugging Face API Key: Wenn man den API-Key eingibt und speichert, wird er beim erneuten Öffnen des Eingabefensters als "n8n_BLANK_VALUE_xxxxx " angezeigt – der ursprüngliche Key ist dann verloren. Lösung: Statt den Key direkt ins Feld zu schreiben, nutze Expression Mode und gib ihn als festen String im JS Format ein: {{ "hf..." }} So bleibt der Key erhalten.

I tried this and it seems to pull the API key correctly, but wont allow me to do anything. i get the "error fetching blob" message whenever running an embedding node

[barl0g]

Workaround für n8n-Fehler bei Hugging Face API Key: Wenn man den API-Key eingibt und speichert, wird er beim erneuten Öffnen des Eingabefensters als "n8n_BLANK_VALUE_xxxxx " angezeigt – der ursprüngliche Key ist dann verloren. Lösung: Statt den Key direkt ins Feld zu schreiben, nutze Expression Mode und gib ihn als festen String im JS Format ein: {{ "hf..." }} So bleibt der Key erhalten.

Excellent! It works!

[schrothbn]

Hello everyone,
The proposed workaround will only skip the credential test but as @jakerios pointed out there are more issues. The underlying problem is that Huggingface updated their API and SDK. Langchain is currently adjusting to it, we will adjust the nodes as soon as they release the fixes on their side. Their PR can be tracked here: langchain-ai/langchainjs#8237

[Oceaneyes123]

I also have same issue. Any fix?

[jakerios]

Nope, looks like we're still waiting on Langchain who has not posted any updates in a while according to the PR linked by @schrothbn above

[jakerios]

The issue I'm facing and what I believe many others are is that I am using the HuggingFace embedding model, and as of right now n8n does not support an HTTP node for embedding. This makes it basically impossible to run a RAG agent with a HuggingFace model as of right now with the current issues. Hopefully this can be resolved soon, its been over a month now facing this issue