# Browse connection data

The connection detail page displays the data surface supported by its connector. Some connections expose a browseable resource; others expose search or tools. An empty data browser can mean there are no visible items, while a connection with tools alone may have no file or record listing at all.

## Files and documents

For a file connection, navigate folders using the folder rows and breadcrumbs. Open a file to read its document view and return to the same browsing context. The viewer can present supported content or a download option when inline rendering is unavailable. Check the key or filename carefully before deleting an object: a delete acts on the provider's stored object.

For a richer multi-provider file workspace, use the [Storage browser](/docs/storage/overview). Importing a file into an App creates a snapshot; simply opening a file from a connection does not.

## Records and tables

A records connector can expose items for inspection and workflow ingestion. A table-backed connection lists tables and opens a selected table's rows on a separate page. Open an addressable row for its individual details, then return to its table context. Load more rows before treating a visible page as the entire table.

## Other surfaces

A vector connection identifies a collection and displays recorded chunk deliveries. Those entries are not an exact inventory of the stored vectors; see [vector deliveries and deletion](/docs/connections/vector-deliveries-and-deletion). A search connection is used by submitting a query. A tools connection offers provider operations that the Assistant can discover and use according to their effects and required approvals.

Browsing original provider data is different from inspecting records already produced in [Data](/docs/data/overview). To answer where an output came from, start with its Data provenance rather than assuming the current provider content is the content used by an earlier run.