List storefront intelligence-run records
Paginated list of intelligence-run records for the storefront, newest first. Each row captures the prompt, raw + parsed LLM output, matched buyer-instructions, and the result envelope. Page with ?skip= and ?take=. Filter the seller demand inbox with ?disposition= (responded / declined_fit / declined_policy / not_live); an invalid value returns 400.
Authorizations
API key or access token
Query Parameters
Number of rows to skip (offset pagination).
0 <= x <= 90071992547409910
Number of rows to return per page (1–100).
1 <= x <= 10025
Demand-inbox filter — return only runs with this brief outcome: responded (the storefront composed a response), declined_fit (declined: no matching inventory), declined_policy (declined: acceptance-policy violation), or not_live (brief arrived before the storefront was live; reserved, not yet emitted). Omit to return every outcome.
responded, declined_fit, declined_policy, not_live "declined_fit"
Case-insensitive substring search over the buyer brief text. Use to find every brief mentioning a vertical, advertiser, or term (e.g. fintech). Combines with disposition.
1 - 200"fintech"
Filter by buying mode — brief for the demand inbox (excludes wholesale/refine runs), wholesale, or refine. Omit to return every mode.
brief, wholesale, refine "brief"
Filter by the commercial result that followed the storefront response, derived from the latest attributed commercial-outcome event: booked (the media buy was forwarded upstream or has reported delivery — a confirmed commitment), pending (the media buy is awaiting the operator’s own acceptance-policy approval and could still be rejected), or rejected (the storefront rejected the media buy). Runs with no attributed media buy yet match no value. There is no lost value — a buyer silently passing on a response has no signal in AdCP today.
booked, pending, rejected "booked"
full (default) returns every run field. summary omits the heavy LLM payload columns (prompt, raw/inventory snapshots) the demand inbox does not render, keeping the response small. The full row is always available on GET /intelligence-runs/:id.
summary, full "summary"