Grafana Llm App

LLM Plugin and frontend libraries repo

This repository holds separate packages for Grafana LLM Plugin and the `@grafana/llm` package for interfacing with it.

They are placed into a monorepo because they are tighlty coupled, and should be developed together with identical dependencies where possible.

Each package has its own package.json with its own npm commands, but the intention is that you should be able to run and test both of them from root.

Contributing

If you're interested in adding support for new LLM providers or extending the functionality of the Grafana LLM App, please see the [CONTRIBUTING.md](./CONTRIBUTING.md) file for detailed implementation guidance.

Developing these packages

Quickstart

1. Install node >= 22, go >= 1.21, and [Mage](https://magefile.org/).

2. Install docker

3. Run `npm install`

4. Run `npm run dev`

5. In a separate terminal from dev, run `npm run server` If you want to bring up the vector services, you can run `COMPOSE_PROFILES=vector npm run server` instead

6. Go to (http://localhost:3000/plugins/grafana-llm-app) to see configuration page and the developer sandbox

This will watch the frontend dependencies and update live. If you are changing backend dependencies, you can run `npm run backend:restart` to rebuild the backend dependencies and restart the plugin in the Grafana server.

Dev Sandbox

It is recommended to develop functionality against the "developer sandbox", only available when the app is run in dev mode, that can be opened from the configuration page, because it gives you a clean end to end test of any changes. This can be used for modifications to the @grafana/llm library, the backend plugin, or functionality built on top of those packages.

Backend

It is recommended to run these using npm commands from root so that the entire project can be developed from the root directory. If you want to dig into the commands themselves, you can read the corresponding scripts in ./packages/grafana-llm-ap