Currently, the AI agent node does not support using API credentials based on headers such as Auth Head, Bearer or even X-API-Key.
Some services such as Chutes.ai, similar to OpenRouter, require header based credentials, and customizing an OpenAI credential to send the API key won’t work.
Of course, we can use the HTTP Request node to send the API key to the platform using a credential previously created, as far as we choose some kind of the headers mentioned before: Bearer token, Authorization Head or X-API-Key.
However, I cannot find any way to orchestrate this HTTP Request so that I can use it from the AI agent node. Even adding the existing HTTP Request as a tool of the agent, it is mandatory to add a model (LLM) that usually requires its own API credential as well.
What I would like is using a language model from the Chutes.ai gateway the same way as we do from OpenRouter or when chosing another LLM in a straight way.
The documentation from Chutes.ai API mentions the following methods as available to receive an authentication request:
API Key in the X-API-Key header
Bearer token in the Authorization header