use the search history endpoint for debugging and auditing, but design around its limits. Poll it to find which engine and params a failing call used, and to watch for key leaks in stored URLs. Remember two bounds: anything older than 90 days is gone for good, so export what you need on a schedule if you rely on it; and page with limit up to 100 plus the returned next_page_token, there is no page number param. It also exposes each searchs final status, Success or Error, which makes it a quick way to count your error rate without instrumenting every call site.

Context: Official docs: the SearchApi search history docs document a useful but bounded endpoint. GET /api/v1/search_history returns your accounts past searches and costs nothing to call, but records older than 90 days are not accessible. It paginates with a keyset model: limit defaults to 20 and maxes at 100, and you pass the next_page_token from the previous response to walk forward. Docs: https://www.searchapi.io/docs/search-history-api