On this page
Pagination and responses
Read collections, filter saved records and download evidence.
Read JSON responses
Successful object responses use { data, meta }. Resource list responses add { pagination: { hasMore, nextCursor } }. The meta object includes requestId and resultScope: saved; reporting endpoints can add their resolved scope or availability. Creation returns 201; successful reads and edits return 200. Dates in stored records use ISO 8601 timestamps. Fields explicitly typed as nullable can contain null. Versioned saved content such as article, payload or research data can have different nested fields depending on the saved content version.
Continue a resource list
Resource lists default to 50 records, with limit from 1 to 200. When pagination.hasMore is true, pass pagination.nextCursor as cursor on the next request. Keep the same key, endpoint, limit and filters. Stop when nextCursor is null. Cursors are opaque and cannot be moved between keys or queries.
Reports can use their own pagination instead: see that endpoint’s page parameters. Crawler events use data.nextCursor as eventCursor and return 25 events per page. Do not apply resource cursor pagination to report rows.
Read the next page
curl --get 'https://app.ansvra.com/api/customer/v1/projects/project_123/prompts' \
--header "Authorization: Bearer $ANSVRA_ACCESS_KEY" \
--data-urlencode 'limit=50' \
--data-urlencode 'cursor=<nextCursor from the previous response>'Choose filters and dates
Use only parameters listed for the endpoint. Resource filters are exact matches. Boolean filters use true or false. Related identifiers must belong to the requested project. Optional archive filters include archived records unless you explicitly pass archived=false.
Supply both ends of an explicit date range. Use YYYY-MM-DD, with at most 366 inclusive days. Resource date filters use from/to; brand reports use start/end and also accept from/to. Historical access begins at the account reporting start. Brand reports resolve dates in UTC. Report-specific defaults and filters are listed on their endpoint pages.
Use platform identifiers
Use openai for ChatGPT, google_overview for Google AI Overviews, google_mode for Google AI Mode, gemini for Gemini, bing_mode for Bing AI Mode, perplexity for Perplexity, claude for Claude, and copilot for Microsoft Copilot. These identifiers can occur in saved history even when new tracking is unavailable for a platform. API access does not enable new tracking.
For markets, use supported uppercase country codes such as US or GB. Language codes are lowercase, such as en or sv. A prompt’s language must match the selected market; omit it on creation to use the market’s language.
Interpret missing information
The Customer API reads Saved results and excludes Overview preview results. It does not collect fresh answers, refresh connected reports or generate content. Missing measurements may be null or unavailable; do not treat them as zero. Rates, comparison cohorts and saved evidence use the same reporting calculations as the app.
Download saved files
Use the documented /download endpoints with the Authorization header. Responses are authenticated attachments, not public file links. Use Content-Disposition for the filename; the content type is application/octet-stream. The maximum file size is 32 MiB. Image files must be available and unexpired; content assets must have a locally saved file. Answer JSON downloads contain normalized saved results. Original response payloads and saved HTML files are unavailable for download. Saved reports also support CSV and PDF; answer downloads support Markdown. CSV report responses use text/csv instead of the JSON envelope.