tool_brave

Brave Search Tool Example

This example demonstrates how to use the Brave Search tool with LangGraphGo to create an AI agent that can search the web using Brave's privacy-focused search API.

Prerequisites

1. Get your Brave Search API key from Brave Search API
2. Get an LLM API key (OpenAI or DeepSeek)

Setup

Set up your environment variables:

export BRAVE_API_KEY="your-brave-api-key"
export OPENAI_API_KEY="your-openai-api-key"
# OR
export DEEPSEEK_API_KEY="your-deepseek-api-key"

Run

go run main.go

How It Works

1. Initialize the LLM: Creates an OpenAI-compatible LLM client
2. Initialize the Tool: Creates a Brave Search tool with custom options:
   • Count: Number of results to return (1-20)
   • Country: Country code for localized results
   • Lang: Language code for search results
3. Create ReAct Agent: Combines the LLM and tool into a ReAct agent
4. Run the Agent: Sends a query and gets the response

Customization

You can customize the Brave Search tool with various options:

braveTool, err := tool.NewBraveSearch("",
    tool.WithBraveCount(10),           // Number of results (1-20)
    tool.WithBraveCountry("CN"),       // Country code
    tool.WithBraveLang("zh"),          // Language code
)

Benefits of Brave Search

• Privacy-focused: No tracking or profiling
• Independent: Not reliant on Google or Bing
• Fresh results: Real-time web search
• Global coverage: Supports multiple countries and languages