Brave Search icon

Brave Search

Search the web using Brave Search

Actions8

LLM Context

AI-generated

Summary

Fetches pre-extracted web content from Brave Search for a required query, formatted as context for AI agents, LLM grounding, or RAG workflows. Configure optional search and context limits to control which sources and how much content Brave includes.

Inputs

  • Query (required) — Search term of up to 400 characters and 50 words.
  • Additional Parameters — Optional settings: country (default US), search language (default English), and spellcheck (default enabled); the maximum search results considered (default 20, up to 50); maximum URLs (default 20, up to 50), total approximate tokens (default 8192, up to 32768), and snippets (default 50, up to 256); context threshold mode (default Balanced); per-URL token limit (default 4096, up to 8192) and snippet limit (default 50, up to 100); one or more Goggle URLs for custom ranking; page freshness (all time, past day, week, month, or year); local recall (default disabled); and source metadata enrichment (default disabled).

Output shape

A single JSON response object from Brave Search for each input item.

The response body is returned as the item JSON. With Continue On Fail enabled, failed items contain the original input JSON and an error.

Links

Discussion