Brave Search n8n integrationAutomate Brave Search with n8n.

Runs a standard web search on the Brave index and returns the result list for the query, ready to loop over in the next node.

• Query: required, the search term, capped at 400 characters and 50 words, often {{ $json.query }} when it comes from a previous node.
• Count: how many results come back, between 1 and 20. Combine it with Offset to walk through further pages.
• Result Filter: restricts the response to the types you want, such as news, videos or discussions. Left empty, every available type comes back.

Same query, narrower index: this operation returns news coverage instead of the whole web, with a much higher result ceiling.

• Query: required, the term being monitored, with the same ceiling of 400 characters and 50 words.
• Count: the number of news results in the response, up to 50. Fewer may come back than requested.
• Freshness: filters by when results were discovered, from pd for the past 24 hours to py for the past year.

Returns image results for a term, with the largest result ceiling of any operation on this node.

• Count: the number of image results, up to 200 in a single call, though the response can be shorter.
• Safe Search: only two levels here, off and strict, unlike the three offered by the web search.
• Language: the search language preference, which changes which sources the images are pulled from.

Video results for a query, with the metadata that lets a workflow decide which ones deserve a human look.

• Count: the number of video results, up to 50, paired with Offset when one page is not enough.
• Freshness: the same discovery window as the news search, useful to keep only clips from the past 7 days with pw.
• User Interface Language: sets the response language, for instance en-GB rather than en-US.

Points of interest in an area. Nothing is strictly required here: leave Query empty and the API returns general points of interest for the area you name.

• Location: a plain-text place. For US cities prefer city, state and country, as in san francisco ca united states, otherwise city and country.
• Latitude and Longitude: decimal degrees, used together as an alternative to Location.
• Radius (meters): biases results toward the coordinates. It is not a strict boundary, and without it the search is global.

Takes a term and checks its spelling, which is the cheapest way to stop a typo from ruining a downstream search.

• Query: required, the term to check, with the usual limit of 400 characters and 50 words.
• Country: an associated country code used as a hint, defaulting to US.
• Language: an associated language code of two or more characters, defaulting to English.

Expands a seed term into the queries people actually type around it, returned as a short list.

• Query: required, the seed term, with a maximum length of 400 characters and a word limit of 50.
• Count: the number of suggestions, minimum 1, maximum 20, default 5.
• Rich Results: enhances suggestions with richer data, and requires a paid Suggest subscription.

Returns pre-extracted web content shaped for AI agents, LLM grounding and RAG pipelines, so the workflow skips the fetch-and-clean step entirely.

• Maximum Number of Tokens: the approximate size of the context, default 8192 and maximum 32768.
• Maximum Number of Snippets: how many chunks of text land in the context, default 50 and maximum 256.
• Context Threshold Mode: how strict the inclusion bar is, from disabled to strict.