Debugging

You have created a JWT LLM Proxy and are ready to use it to proxy your LLM API calls. But when you try to call the proxy you get an error. What do you do?

1. Make sure your LLM API key is properly configured

Call the API directly to make sure your private LLM key is properly configured using curl or a tool like Postman or Bruno. For the latter we have a predefined collection of requests defined for OpenAI, Anthropic and Gemini in our GitHub repo.

2. Make sure your JWT is valid and properly generated

Create a test user with your authentication provider using email and password authentication. Then use the test user credentials to generate a fresh JWT using our generator tool for Supabase or Firebase. Now try to call the LLM API via the JWT proxy using the generated token instead of the private LLM key, and the JWT proxy URL instead of the LLM API URL.

3. Make sure your LLM SDK of choice is properly configured

Backmesh JWT LLM Proxies are compatible with any LLM SDKs and app frameworks. Make sure your LLM SDK is properly configured to let you override the base URL of the LLM API with that of the Backmesh JWT proxy. Check out our tutorials for examples on how to do this for different frameworks:

• Flutter app using Firebase Authentication and Google Gemini API
• Javascript web app using Supabase Authentication and OpenAI API