My use case:
I’m trying to integrate Claude 3.7 Sonnet (anthropic.claude-3-7-sonnet-20250219
) into an n8n
workflow using the AWS Bedrock invoke
endpoint directly:
POST https://bedrock-runtime.us-east-1.amazonaws.com/model/anthropic.claude-3-7-sonnet-20250219/invoke
This model is visible and working in my AWS account — I can access it via the AWS Console and run requests. However, it doesn’t appear in the model dropdown in n8n’s built-in Bedrock node.
Since I couldn’t select it natively, I tried using the HTTP Request node to call the endpoint manually. That’s where I ran into a second problem: the request needs to be SigV4 signed, and the HTTP Request node doesn’t support generating these dynamic headers.
I considered generating the Authorization
header manually, but it’s tied to each request’s body, date, region, etc., and needs to be signed using the AWS secret key — so you can’t hardcode it.
I’m now working around this by setting up a small Lambda function to sign and forward the request, but it would be great to handle this entirely within n8n.
Any resources to support this?
- AWS Signature Version 4 signing process
- Claude 3.7 Sonnet on Amazon Bedrock – official release
- API Endpoint:
https://bedrock-runtime.us-east-1.amazonaws.com/model/anthropic.claude-3-7-sonnet-20250219/invoke
Are you willing to work on this?
Yes — if there’s any guidance on how to extend the Bedrock node or integrate SigV4 signing in a Function node, I’d be happy to help build or test a solution.
Also, is the list of models in the Bedrock node hardcoded or cached? I have access to Claude 3.7 Sonnet in my AWS account, but it’s not visible in n8n.
Would really appreciate any insights or suggestions!
Thanks,
Chetan