Local Indexing

[sick vector]

Omg... how to setup local indexing on my homelab server ? i connect via ssh to server configure everything etc and always have this info on screen, on my local pc its fine i don't do nothing and it worked i can change all indexing setting.

[round arrow]

Thats not local indexing, that is Managed Indexing in the Kilo Cloud,

Local indexing is here. Click on the little cylinder with the dot (yours will not be green).

That entire scrollview is the Settings for it.
https://kilo.ai/docs/features/codebase-indexing

All you need to have is an API key for OpenRouter or OpenAI, pick an embedding model (I use Qwen 8B on OR) and you can use LanceDB (no setup - free) or Qdrant (local Docker setup - free) for the vector database.

Any questions I am around

[sick vector]

yeah i figure it out, needed at start remote connected setup indexing then it start work, damn and i sit on it since hour figure out config files etc XD

This cylinder changed on that from my screen when i go to workspace folder without setup indexing

anyway it worked 😄

thx for answer 🙂

ps. im not programmer just trying new things that's why this was hard for me 😄

[round arrow]

NP... we are a friendly bunch here. Ask all the questions you want! ❤️

[olive bear]

Sir, is the codebase index feature available for using local models (OpenAI Compatible)? I'm having trouble using local models; I keep getting an error message saying the URL endpoint is wrong (I'm sure my endpoint is correct; I used that URL endpoint in the model agent configuration in Kilo Code settings).

[olive bear]

@round arrow Sir, i need your help 😓

[round arrow]

I am AFK so I can't test, but try adding a /V1 to the end of the URL for OAI. Some local LLM's enginges expect / expose one of them.

Edit: It could be /v1, /openai/v1 OR /v1/openai depending on what you are talking to. 🤷

[round arrow]

And remmber you need an embedding model. A normal LLM will not work very well at all.

[olive bear]

Sir, is it possible for me to use Docker Model Runner with a local QWen3/embedding model for codebase indexing?