BrowserStack Automate REST API – Pagination Consistency Across Dataset Changes

The Reality of Offset-Based Pagination

The BrowserStack Automate REST API utilizes offset-based pagination, which is inherently stateless. Because there is no cursor or snapshot mechanism, the API does not provide a point-in-time view of the dataset. If the underlying data changes between two requests, the index will shift, leading to either duplicate entries or missed records.

Addressing Unresolved Behavior

• Duplicate IDs: The API does not guarantee the avoidance of duplicates. If new test runs are completed and inserted at the beginning of the list while you are paginating, the entire set shifts forward, causing the last item of your previous page to appear again on the current page.
• Data Gaps: If test runs are removed or deleted between paginated calls, the dataset shifts backward. This causes the offset to skip over a record, resulting in a gap in your collected data.
• Maximum Limit: While not explicitly documented, most high-scale APIs enforce a hard cap on the limit parameter to prevent database exhaustion. Exceeding this typically results in silent truncation to the maximum allowed value rather than a 400-series error.

Strategies for Consistent Retrieval

Since the API cannot guarantee consistency across requests, you must implement client-side logic to ensure a complete set:

1. Client-Side De-duplication: Store all retrieved Test IDs in a Set or Hash Map. As you iterate through pages, discard any ID that already exists in your collection to handle duplicates.
2. Filtering by Timestamp: If the API supports a since or created_after filter, use it to narrow the dataset to a specific window. This reduces the surface area for new records being inserted during the enumeration process.
3. Iterative Reconciliation: For critical data, perform a final pass with a high limit or a specific filter focused on the most recent records to capture any data that might have shifted during the initial crawl.

To provide a more specific recommendation regarding the limit cap, are you attempting to fetch thousands of records in a single request, or is your primary concern the frequency of test completions?