Full detail for one resolved value, including the resolved value
itself, the resolution status and justification, the dimension
instances it belongs to, and the document pages it was sourced from
(with web supplement data when web search is enabled on the agent).
Identifier of the underlying atomic value (one per combination of agent and dimension values).
result_name
Yes
string
Human-readable label for this result, typically a concatenation of its dimension values.
agent_name
Yes
string
Name of the agent that produced this result.
value
Yes
any
Resolved value. Shape depends on the agent’s value_type: string -> string; number -> number; bool -> boolean; date -> ISO 8601 date string YYYY-MM-DD; datetime -> ISO 8601 date-time string. May be null for missing values.
Citation map keyed by reference_id (the same IDs referenced from justification_for_resolution_status and from extraction passages). Each value is a 2-element array [document_name, document_page]: document_page is the 1-based page number, or null for sources without pages (e.g. websites). Empty object {} when there are no citations.
Where the resolved value came from. document when the value was derived from uploaded documents only; web when it was derived from web-search results only; document_and_web when both contributed. The non-document values only appear when enable_web_search is set on the agent.
is_web_supplement
Yes
boolean
True when this result was supplemented with web search data rather than derived purely from uploaded documents.