Create a source from a connection
Set up manual or scheduled collection with a reviewed source mapping.
A saved connection authorises access to a service. A data source defines which information to collect from that connection and how to represent it in Strive. Creating the connection alone does not establish a recurring collection.
Ask the Assistant to set up collection
Open the Assistant and describe the connected service, the collection you want, and the refresh frequency. For example: “Use my research documents connection to collect the project notes as a data source named Project research. Show me the mapping before enabling hourly refresh.”
The Assistant inspects the service’s available tools or catalogue ingestion configuration and proposes a collection plan. Review the proposed connection, name, listing and fetch operations, record mapping, and schedule before approving. Use an existing connected account; enter credentials only through its connection form.
Omit a schedule for manual-only collection. Use a recurring schedule only when you want ongoing ingestion, and confirm its timezone rather than assuming your profile timezone applies. A one-off question can use a provider’s tools directly without creating a retained, scheduled data source.
Review the collection mapping
| Setting | Purpose | What to check |
|---|---|---|
| Connection | Identifies the authorised service account | Correct account and resource access |
| Display name | Names the source in Data | Recognisable and distinct from other sources |
| List operation | Finds the candidate records | Appropriate scope and supported tool |
| Records field | Locates a records list in a wrapped response | An array path such as $.items, rather than a scalar property |
| Cursor field | Locates the continuation value | All intended pages can be collected |
| Fetch operation, when needed | Retrieves each listed item’s full content | Correct item identifier argument |
| External identifier mapping | Preserves source identity | Stable identifier, not the position in a page |
| Title mapping | Provides a readable record title | Useful text even when several records have similar names |
| Content text mapping | Provides searchable document or record text | The actual content, rather than only a link or empty preview |
| Source-updated mapping | Carries a provider modification value | Compatible values from the selected record shape |
| Sync schedule | Sets recurring ingestion; absent means manual | Intended frequency, timezone, and processing volume |
Mapping paths select values from the provider’s response. A list operation can return a wrapper containing both records and a next-page token; the records path and cursor path serve different purposes. Some services list only metadata and require a separate fetch to retrieve content. The proposal should follow the discovered service contract, not guessed operation names.
Verify the first collection
After approval, open Data and find the named source. Start a manual sync where available and inspect sync history. Compare collected count, titles, text, source identifiers, and provenance with a small known set in the provider.
An empty successful collection can mean the selected scope contains no matching records. Missing later pages can indicate a pagination mapping problem. Blank text can indicate that a fetch or content mapping is wrong. Review those boundaries before broadening access or scheduling repeated runs.
Maintain or remove the source
When provider tools, response shapes, permissions, or the intended scope change, review the collection configuration again through the Assistant. A reconnect restores account access but does not automatically correct an incompatible mapping. Inspect a successful updated sync before relying on recurring runs.
Deleting collected source records and deleting the saved connection have different effects. Read source deletion and connection lifecycle before removing either. Preserve needed exports separately from an app definition.