Store
Organizing the Jobs' outputs across time and entities
The Store is Deck’s persistent memory. It captures and organizes everything jobs produce — not just during a single session, but across all sessions and all connections related to an end-user or entity.
It’s where developers retrieve the actual value generated by their jobs.
Structure of the Store
- Entity-centric: All data is indexed by the real-world entities that Jobs act on (e.g. accounts, locations, meters)
- Time-aware: Every output is timestamped and stored with Job lineage
- Job-aware: You can query data from a specific Job or across Jobs of a given type
Why it matters
- Stability across connections: Even if credentials change or expire, data remains linked to the same account or entity
- Structured access: Developers can query the Store via simple API endpoints — by account, Job type, or time range
- Enables product logic: Power timelines, dashboards, monitoring, history comparisons, and more
What lives in the Store
- Documents — statements, confirmations, exports
- Normalized records — balances, usage, payments, statuses
- Messages — confirmations, error logs, or audit trails
When to use the Store
When an agent extracts data from a website, the usual endeavor is to return said data as JSON via a webhook. However, there are cases where the amount of data extracted exceeds what a webhook can return. In these cases, the data will be written in a CSV file, then stored in our Store.
The webhook will return a URL which has already been filtered so you can access only the document that was created with the data you wanted at the endpoint /jobs/documents/list. Using that URL will give you a list of documents, each with a document ID, which you can then download from the endpoint /jobs/documents/file.
Updated 11 days ago