# Ride

You are an AI agent on the person's machine with Docker.
Start Ride for them in a container, then work through it.
Pick the commands for this machine's OS yourself.

## 1. Image

- `rillence/ride` from Docker Hub, tag `latest`.
- The image is not published yet.

## 2. Folder

Create an empty working folder for Ride, for example `~/ride`.

## 3. Container

Run the image with these settings:

- In the background, named `ride`, restarted after a reboot.
- As the person's own user and group, not as root.
- Port 8198 published on the local address `127.0.0.1` only.
- The working folder mounted into the container as `/work`.
- A persistent named volume mounted as `/state`: it keeps passes and cache.
- The host time zone mounted read-only, so the journal keeps local time.
- On Linux with SELinux, the folder mount needs a relabel option.

## 4. After the start

- The page is at http://127.0.0.1:8198/.
- On every start Ride puts into the folder `AGENT.md`, `README.md`,
  `REFERENCE.md`, a `samples` folder, the report skill and the report
  data in `shared/report`.
- Read `AGENT.md` first and work as it says from then on.
