What is Kernel Search?
Kernel Search is the search-specific crawler operated by Kernel Technologies, Inc. It is separate from Kernel Agent, which runs browser automation in response to an end user's request. Kernel's bot documentation gives each identity a different purpose and verification authority.
The search crawler builds indexes and retrieval databases from web pages. Content it can fetch may become available to Kernel-powered search or retrieval, although a crawl does not guarantee that a page will be selected, quoted, or ranked for any query.
Kernel describes this identity as a search bot, not a model-training collector. There is no published claim that a Kernel Search visit adds the fetched page to foundation-model training data. Its direct consequence is eligibility for Kernel's search and retrieval systems.
Requests carry the KernelSearchBot token. Cloudflare's bot directory lists the full user agent as KernelSearchBot/1.0 (+https://www.kernel.sh/docs/bots), which is useful for log classification but can still be copied by an unrelated client.
Kernel also signs Search requests with Web Bot Auth. The Search identity publishes its keys at https://search.bot.kernel.sh/.well-known/http-message-signatures-directory. A verifier can use the Signature-Agent, Signature-Input, and Signature headers to check the request instead of accepting the user agent at face value.
Robots support is not settled in the supplied records. Kernel says the crawler follows rules for KernelSearchBot, including crawl-delay preferences, while Cloudflare's bot directory marks it as not following robots.txt. This page therefore treats compliance as unknown and recommends checking actual requests after a rule changes.
