No AI nodes have proxy support?

[sac.mm.xlv]

There’s no way to specify a proxy in any of the AI nodes. Am I limited to HTTP nodes if I’m behind a proxy?

[jabbson]

Hey @sac.mm.xlv hope all is well and welcome to the community.

Since you didn’t specify it explicitly, I will assume you are talking about self-hosted version of n8n.

If you are running your n8n from behind a firewall and need to setup a proxy this can be done with setting up environment variables, HTTP_PROXY and HTTPS_PROXY.

See if this thread helps:

Http proxy problem

having a universal proxy toggle would be handy

[sac.mm.xlv]

Hi thanks for the welcome we are testing n8n locally, and will self-host a community edition before deciding to switch to enterprise/hosted.

We give the container access to the host’s network:

docker run \
-it \
--rm \
--name n8n \
-p 5678:5678 \
--network="host" \
-v n8n_data:/home/node/.n8n \
docker.n8n.io/n8nio/n8n

The HTTP node has a proxy field and works fine, but the AI nodes don’t. Does this mean an environment variable (hence enterprise version) is needed to use the AI features with a proxy?

[jabbson]

@sac.mm.xlv

Does this mean an environment variable (hence enterprise version) is needed to use the AI features with a proxy?

environment variables do not require you to run enterprise version of n8n. You can have environment variables passed into docker container in a self-hosted configuration.

[sac.mm.xlv]

Ok thanks makes sense. Just to confirm, if we want shared env vars (proxies, API keys etc) in a self-hosted config then they need be configured as n8n variables, and this does require enterprise? But these variables are securely hidden (can’t be leaked or visible) when building workflows?

[jabbson]

I guess this depends on what environment we are talking about. http_proxy and https_proxy are system variables - you give these values to the system to use when a system networking call is made. On the other hand, environments that you were mentioning (the one that is a paid feature) is used to define workflow context (e.g., “production”, “staging”) for different setups. For each such environment you can configure things like workflow variables and secrets.

As to how secure things are - there is never a guarantee that your data is secure, As my professor used to say - “the most secure system is the one that was never built”.