Keep on getting the “credentials not found” error when working with http node as a tool for the AI agent I use. I have a stand-alone http node which perform exactly the same task, with the same credential set up for the url endpoint calling. Yet, it seems like the one with AI agent does not work at all. Keep on hitting the error. Any recommended fix?
It looks like your topic is missing some important information. Could you provide the following if applicable.
- n8n version:
- Database (default: SQLite):
- n8n EXECUTIONS_PROCESS setting (default: own, main):
- Running n8n via (Docker, npm, n8n cloud, desktop app):
- Operating system:
Hi @ria I am currently running on ver 1.71.2
Hi @Yin
Can you check in the logs of the agent for more details on the error from the HTTP request node? Also, please verify that you are using the latest version of the HTTP tool node (see the settings tab).
Hello,
I’m experiencing the exact same problam as described by @Yin : same request works great in the standalone HTTP node, but it fails in the HTTP Tool.
Here is my configuration:
- n8n version: 1.76.1
- Database (default: SQLite): default
- n8n EXECUTIONS_PROCESS setting (default: own, main): default
- Running n8n via (Docker, npm, n8n cloud, desktop app): Docker
- Operating system: MacBook Pro M1
- HTTP Request Tool node version: 1.1 (Latest)
There is no further information on the AI Agent logs; this is all I see:
{
"response": "There was an error: \"Credentials not found\""
}
Any clues?
Thank you.
I think it may be related to the issue I described here
For me the HTTP Request tool seems to work with predefined credential type but not with generic ones
I’m using version 1.76.1 and latest version of the node.
Agreed; they are related.
Unfortunately I can’t figure out know how to use the Predefined Credential Type with a Keycloak identity Provider.
Issue still exists on version 1.76.2. Using generic credential (oauth2 api) with http node works. when using the http tool call node it doesn’t
Same issue. Problem with executing an HTTP Tool node with a generic credentials type (OAuth2 API). Is there any workaround for the case? I need to perform an HTTP call by Agent and get a result
The workaround is to use Call n8n Workflow Tool with additional workflow and a regular HTTP request node. Works good
Hi ria,
looks like issue still exists in v 1.80.5 cloud. The work around is not ideal and in some cases not a workaround at all. Any potential solution on the horizon ? This is very important for us.
Hi guys,
we’re looking into it!
Thanks
Thanks Ria! I’m glad to have stumbled into this thread and seen the workaround above since I was about to post about this exact issue as well.
The workaround is not ideal if you have an agent with 10 HTTP request tools. This means that you now have to create and manage 10 different workflows. Moreover, you cannot leverage the main functionality where you want the model to decide the query parameters which then makes the work around not feasible.
You are right. This solution smells. However, I solved the trouble with a lot of different requests by creating a variable in which I send from the agent tool a type of request. Then use a switch to decide which endpoint to call. It is really bad thing but no other way to solve this
Hi ria, any news on this? … Many thanks.
I could handle this issue and make by using subworkflow tool and there executing the standard HTTP node with the request
Hi all, we’re working on a new version of the HTTP Request Tool. It’s taking a while though… Since there’s a (albeit cumbersome) workaround, if you run into this issue, please use the subworkflow tool and a normal HTTP Request Node instead until we have this sorted.
Thanks for the update, jpvanoosten. Any idea on when is the release planned?