Brand Alignment: source history

GET /v1/brands/:brand_id/alignment/sources/history

Access is limited to the brands permitted by your credential. Requires read:alignment. Workspace API keys and signed-in MCP sessions are supported.

How one cited page, or every cited page on one domain, shaped what AI says about your brand across finished scans, oldest first, as the “Influence over time” chart in a source’s sheet shows it: up to 26 scans ending at scan_id. Give url or domain.

Asky manages the monitoring questions. Responses include results without exposing generated question text.

Query parameters

Parameter	Type	Description
`url`	string	One cited page, as the sources endpoint lists it. Give url or domain.
`domain`	string	Every cited page on this domain, such as example.com (www . and m. read as the same site). Give url or domain.
`scan_id`	uuid	The last scan to include, from the program endpoint. Defaults to the latest finished scan.
`models`	string (repeatable)	Engine identifiers. Repeat this parameter to select multiple engines.
`markets`	string (repeatable)	Country codes (`DE`) or one country and language (`DE:de-DE`). Repeat this parameter to select multiple markets.
`stream`	string	`dedicated` for answers to Asky’s scan questions, `ordinary` for tracked-prompt answers that mention the brand, or `all` (the default).
`since`	date	Only scans on or after this date (YYYY-MM-DD).
`by_engine`	boolean	Also the counts per engine.

Response

The standard API envelope places the result in data, with request_id beside it at the top level.

Field	Description
`source`	The `url` or `domain` asked for.
`scans`	Oldest first. Scans with no answer counted in the selection are left out.
`scans[].contradicted_claims`	The AI claims about your facts that got your brand wrong and trace back to the source: `count`, and the `answers` holding them. Each claim counts once, however many facts it addresses; a domain counts each once across its pages.
`scans[].aligned_claims`	The same, for AI claims that got your brand right.
`scans[].citation_rate`	The share of the answers counted that cite the source (`value` in percent with its `numerator` and `denominator`). Each answer counts once.
`scans[].mentions`	The mentions of your tracked attributes that trace back to the source, by sentiment (`positive`, `mixed`, `negative`, `neutral`): each passage of an answer that ties an attribute to your brand counts once, in its own sentiment, with your team’s corrections; a domain counts each once across its pages. Where no check ran on an answer, the engine’s citation next to the passage stands in, as it does for the AI claims.
`scans[].change`	Since the scan before: the difference in contradicted and aligned claims, and the citation rate’s change in percentage points. Null for the first scan only; given across a setup change too.
`scans[].setup_changed`	True where the monitoring setup differs from the scan before. The chart keeps the line running and marks the scan.
`scans[].by_engine`	With `by_engine`: the same counts per engine, for the engines with answers counted.

The cited pages of one scan are the sources endpoint. Scan identifiers come from the program endpoint.