# Install

This page says what AI-LinkMO needs on the machine, how to install it, and how to start the API, the web UI and the CLI. [Working with the data](https://noelmcloughlin.github.io/ai-linkmo/docs/working-with-the-data/) says what to do next.

> **Only want to ask?** Install `ktl-docent` into any agent you use, by a route from [the install page](https://w3id.org/ktl/docs/install/), and ask it about this project, such as *where the EU AI Act encoding comes from, and who checked it*. It answers from the bundle and says how far each answer has been checked.

## Install

AI-LinkMO needs Python 3.11+ and [uv](https://docs.astral.sh/uv/), and Node 22 for the web UI.

```bash
uv sync --extra test
./ai --help
```

That is the offline CLI, ready to use. If `uv sync` needs to compile anything on your machine, or you have upgraded `ai-atlas-nexus` and need to regenerate the UI's schema, see [docs/install-notes.md](https://noelmcloughlin.github.io/ai-linkmo/docs/install-notes/).

## Quick start

1. **Start the API** ([details](https://github.com/noelmcloughlin/ai-linkmo/blob/0d8d49f45be30e5f8e9d661d9edeacba9446570a/lib/api/README.md)). It listens on port 8000. Its write endpoints are not authenticated, so keep it on localhost.

   ```bash
   uv run uvicorn lib.api.server:app --reload
   ```

2. **Start the web UI** in a second terminal ([details](https://github.com/noelmcloughlin/ai-linkmo/blob/0d8d49f45be30e5f8e9d661d9edeacba9446570a/lib/frontend/README.md)). The login in the top corner is a persona picker for the demo, not authentication. **Curate mode**, which edits records in your own data files, is reached through it.

   ```bash
   cd lib/frontend
   npm ci
   npm run dev
   ```

3. **Try the CLI** in a third terminal. Each entity type has its own command, and [docs/cli-examples.md](https://noelmcloughlin.github.io/ai-linkmo/docs/cli-examples/) lists them all with the flags they share.

   ```bash
   ./ai -h
   ./ai risk -h
   ```

[docs/working-with-the-data.md](https://noelmcloughlin.github.io/ai-linkmo/docs/working-with-the-data/) says how to go on from there: bring your own data, compute a crosswalk, search the records, browse them in Datasette, or load the graph into Neo4j.