The ai does not respond after Intranet penetration

[sizukamasou]

I have successfully deployed the node-chatgpt-api and pandoraAI.

And I can run it on my computer. Talk to bing normally.

I then mapped the access address 192.168.0.101:3000 to a URL using ngrok.

My other devices can access this mapped URL, but cannot interact with the ai.

I don't know how to set it up so that I can access my deployed pandoraAI from another device or location.

Here is my configuration:

I'm using clash's global proxy, ports 7890

Similarly, I've deployed pandora in the past，It is chatgpt3.5. It works fine with the same through-mapping method

Please help me, thank you!

[waylaidwanderer]

Make sure you're not running the client in dev mode if you're accessing it from outside. You have to build it and run the production build instead.

[waylaidwanderer]

Oh, and the API_BASE_URL needs to be behind Ngrok as well since the client will be connecting to this URL client-side.

[sizukamasou]

哦，API_BASE_URL 也需要在 Ngrok 后面，因为客户端将连接到此 URL 客户端。

Oh, and the API_BASE_URL needs to be behind Ngrok as well since the client will be connecting to this URL client-side.

Still no success I ran npm run buildmingl as you suggested，The first time I opened the page, it showed 500 errors.

Then chatgpt told that it was missing the configuration of setting.js under pandoraAI,Originally I couldn't find it on github, so I asked gpt to tell me what was in it, and I tried to create it

(https://7v4x539662.imdo.co/It is the domain name mapped by ddns)

then

I tried this domain name in api_base_url under the.env file I also tried using http://localhost:3001

And shut down the firewall on the computer I then successfully started npm run build

I can then communicate with the ai through this domain name

But the phone still couldn't interact

I really don't know how to change my configuration,Or is there a problem with my ddns software?

[waylaidwanderer]

This isn't a bug with the repo so I'm converting this issue to a Help thread.