What is KimiBot?
KimiBot is Moonshot AI's crawler for content that may be used to train Kimi foundation models. Moonshot documents it on the Kimi Crawlers policy page, alongside separate agents for search and user-requested retrieval.
Those roles are independent. Kimi-SearchBot analyzes pages and builds the index behind Kimi search, while Kimi-User fetches pages for actions such as summarizing a particular article. KimiBot is the training crawler. Blocking it does not, by itself, block the other two workflows.
A KimiBot hit shows that a URL was fetched for possible training use. It does not confirm that the page entered a dataset or influenced a model. KimiBot also has no documented role in live Kimi search, so the visit is not evidence that a page will appear or be cited in a Kimi answer.
Cloudflare's bot directory records the full user agent as Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; KimiBot/1.0; +https://www.kimi.com/policies/kimi-crawlers. Moonshot also publishes current KimiBot IP ranges from its policy page. This bot entry is not IP verified and has no Web Bot Auth signature directory, so the token alone should not be treated as authentication.
The robots.txt evidence is mixed. Moonshot says KimiBot uses industry-standard directives and gives examples for full-site blocks, path exclusions, and Crawl-delay. Cloudflare's directory marks the crawler as not following robots.txt, while the bot's listing leaves the status unknown. That conflict calls for log verification rather than a categorical promise.
Sites that accept possible use in Kimi model development can leave KimiBot open while still protecting private routes normally. Publishers that want Kimi search discovery without training collection can block KimiBot and set separate rules for Kimi-SearchBot. Each subdomain needs its own robots.txt policy according to Moonshot's documentation.
