A reference tells you what the requests are. A collection is a thing you
**open** — so every Bruno section offers the files themselves.

## The archive

The overview links a `.zip` built during the build from the exact version and
language of the page it sits on. Nothing is configured; it is simply there.

It contains the collection as committed, **minus `environments/`** — see
[what is published](/reference/bruno/what-is-published).

> [!NOTE]
> A collection in a **downloaded** repository gets no archive. Content's
> checkout lands in a hash-cached directory known only inside the collection
> that declared it, so nothing outside can read the files a second time. The
> overview then links nothing rather than linking a file that was never written.

## Fetch in Bruno

`options.fetch` takes a public git URL and adds a **Fetch in Bruno** action
beside the download:

```ts
options: { fetch: 'https://github.com/acme/api.git' }
```

It is a plain outbound link to Bruno's own deep link. duxt does not load the
`button.js` Bruno publishes for it — a third-party script on every page of a
documentation site, to render an anchor the site can render itself.

> [!IMPORTANT]
> It is **not a pinned substitute for the archive**. It clones the repository's
> default branch, which is today's collection and not the version on screen. The
> page says so, and offers both rather than presenting them as the same thing.