Instructions for self hosting

[jaycangel]

There don’t appear to be any clear instructions for self-hosting in the current documentation.

FORGE_WORKSPACE_SERVER_URL is mentioned, but there’s no explanation of what it represents, how it should be configured, or what value should be used in a self-hosted setup, or what other packages or software should be installed.

It would be helpful to include:

• A brief explanation of what this variable controls
• Whether it’s required for self-hosting
• An example configuration for a local or self-hosted environment

At the moment, it’s difficult to understand how to proceed with a self-hosted deployment based on the available docs.

[bkutasi]

llama.cpp integrated seamlessly for me. Do you have any problems hooking up forge to your endpoint?

[jengel7777]

any hints on what i need to setup? what model do you have to install if you are on LM Studio on mac what mlx model can you install? i just wasn't sure when it said self host what it was meant to point to? i assume it's some type of encoder and vector store?

Do you host just the LLM or do you host the FORGE_WORKSPACE_SERVER_URL - the full indexing workspace?