An agent retrieving information needs more than a successful response. It needs to know whether there is more to read, whether the requested scope was available, how much the request used, and what to do when the answer cannot be completed.
Our latest API improvements make those decisions clearer—for the agent doing the work and the person relying on its answer.
Keep going beyond the first page
Research rarely fits into one page. Stream listings and individual Stream Manifest feeds now return fully qualified next and previous URLs. An agent can fetch the next URL with the same authentication while keeping the selected filters and page size. When there is no next page, the response says so explicitly.
That means an integration can work through the available results instead of quietly treating the first page as the complete picture.
POST queries stay explicit
Advanced query and execution requests use a different contract: they remain POST requests. Agents preserve their original scope and use the pagination values to request another page. Planned execution requires a fresh plan token when changing pages.
The distinction matters. A paging URL on these POST endpoints is a precise pointer for the next request; it is not a link an agent should blindly GET.
Know when “nothing returned” is not “nothing exists”
An empty answer and an unavailable answer are different. If a count request cannot be authorized for the selected Stream, Synorb reports that the count is unavailable—not an exact zero. An agent can explain the limitation instead of telling its user that no matching information exists.
Query responses also expose structured information about coverage, scope authorization, citation expectations, and suggested corrections. These give integrations more context for deciding whether to present an answer, clarify the request, or explain a limitation.
See what each retrieval actually used
Query and execution receipts report the Manifests billed for that call alongside the updated quota balance. That gives an agent a usable checkpoint after retrieval: it can tell its user what the request consumed and make its next decision from the current balance.
Recover with a clear next step
Invalid-key responses provide a recovery action and destination. Missing-route responses point callers toward the documentation and machine-readable API reference. For builders, recovery is easier to implement; for users, there are fewer unexplained dead ends.
These changes make Synorb easier to build into research assistants, recurring monitors, and source-backed applications: clearer navigation, clearer limitations, and clearer usage.
Read the agent and API documentation or start building with Synorb.