I'm trying to create a GPT that can access my Localhost for use with an API. It for some reason cannot do this, however I am able to access the API landing page via my browser. This used to work with ChatGPT Plugins but I'm not sure on what to do.
#GPT cannot use Localhost for Actions
whole condor
brave nimbusBOT
Hi @whole condor. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Based on the information from the OpenAI community forum, it seems that GPT Actions may not directly support localhost development. However, there are ways to work around this.
One suggested method is to use a service like ngrok.io (http://ngrok.io) to create a Cloud Edge, which will provide you with a URL that you can use in the openapi spec. You can then run ngrok locally to create a tunnel from the ngrok server to your localhost. Here's an example command:
./ngrok.exe tunnel --label edge=edghts_fjsiasuhjkerwgjZMw <http://localhost:80>
This command will create a tunnel from the ngrok server to your localhost on port 80. The edge=ed[…] value would be unique to your account/endpoint. Note that you'll need to do an authorization step in ngrok first.
This method also downgrades from HTTPS, which OpenAI sends requests in, to simpler HTTP, which might be easier to receive. However, there might be safety considerations, and you can probably leave it as HTTPS if you prefer.
Another potential issue could be related to the certificate chain setup on your server. If your server doesn't provide a certificate chain, GPT's code to access your API might block the responses. This issue was reported by a user who had their load balancer on AWS set up without a certificate chain.
Please refer to the original forum posts (https://community.openai.com/t/538074) for more details.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
whole condor
@brave nimbus ok thanks i will try this