What is MistralAI-User?
MistralAI-User fetches a page when someone asks Le Chat a question that calls for current web information. Mistral can use the retrieved material in its answer and include a link to the source. The user request is the reason for the visit.
This identity is not Mistral's automatic web indexer. It retrieves pages in real time for a product action, rather than roaming the web to maintain a corpus. Logs may therefore show a focused request to a page relevant to the question instead of an extended site crawl.
Allowing MistralAI-User lets Le Chat read a current page at answer time. Blocking it can prevent that live retrieval and any source link that depends on it. A rule for this token does not, by itself, control the separate MistralAI-Index crawler or remove material that another search system already holds.
Mistral says this user-action bot is not used to crawl content for generative AI training. Blocking it is therefore not a training opt-out. Mistral publishes MistralAI-Training as the distinct identity for collection used to build model training datasets.
Robots handling needs a qualified answer. Mistral's operator documentation presents MistralAI-User as the tag that governs where user requests can go, but this catalog marks compliance as partial and Cloudflare's bot directory marks followsRobotsTxt as false. A directive is useful, though it should not be your only barrier.
The full recorded user agent is Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; MistralAI-User/1.0; +https://docs.mistral.ai/robots). Promptwatch tracks the bot and marks its IPs as verified. Cloudflare's bot directory lists the same user agent and no Web Bot Auth directory, so validation depends on the token and Mistral's published network information rather than a request signature.
