brave-search

Use when searching the web via Brave Search API, finding images, news, or getting AI-grounded answers with citations. Triggers on "search brave", "brave search", needing current web info, image searches, news lookups, or requests for cited AI answers.

⭐5stars🔀1forks📁View on GitHub🕐Updated Jan 11, 2026⬇️2downloads

When & Why to Use This Skill

This Claude skill integrates the Brave Search API to provide high-performance web and local search capabilities. It enables AI agents to access real-time information, discover local businesses, and retrieve raw search results to provide grounded answers with accurate citations, bypassing the limitations of static training data.

Use Cases

  • Real-time Information Retrieval: Fetching the latest news, technical documentation, or current events to ensure AI responses are up-to-date.
  • Local Business Discovery: Searching for restaurants, services, or shops nearby, including details like addresses, ratings, and phone numbers.
  • Fact-Checking and Citations: Verifying claims by searching the live web and providing direct links to primary sources for enhanced credibility.
  • Market Intelligence: Conducting targeted searches for competitor analysis, industry trends, or specific data points across the web.
namebrave-search
description>
allowed-toolsBash, Read
short-descriptionWeb + local search via Brave API

Brave Search

Web and local search using the Brave Search API. Returns raw results (not LLM-summarized).

Prerequisites

  • BRAVE_API_KEY or BRAVE_SEARCH_API_KEY in environment or .env
  • Install CLI deps: pip install typer

When to Use

  • You need raw web results without LLM synthesis
  • You want local business info (addresses, ratings, phone numbers)
  • You want a second opinion vs other search tools

Quick Start

# Web search (JSON by default)
python .agents/skills/brave-search/brave_search.py web "site:openai.com gpt-4o"

# Local search
python .agents/skills/brave-search/brave_search.py local "coffee near Cambridge MA" --no-json

CLI Usage

python .agents/skills/brave-search/brave_search.py web "query" [--count N] [--offset N] [--json/--no-json]
python .agents/skills/brave-search/brave_search.py local "query" [--count N] [--json/--no-json]

Python API

from brave_search import web_search, local_search

results = web_search("site:openai.com gpt-4o", count=5)
local = local_search("pizza near Boston", count=5)

Agent Tool Usage (MCP)

If MCP tools are available, prefer:

  • mcp__brave-search__brave_web_search for general web queries
  • mcp__brave-search__brave_local_search for places/nearby queries

Examples

python .agents/skills/brave-search/brave_search.py web "ArangoDB ArangoSearch BM25"
python .agents/skills/brave-search/brave_search.py local "restaurants near Pike Place Market" --no-json

Tips

  • Use --no-json for quick human-readable output
  • Local search falls back to web if no locations are found