brave-search

Purpose

This skill provides access to Brave Search, a privacy-focused web search API that enables querying the web without tracking, while supporting filters and snippet extraction for AI-driven applications.

When to Use

Use this when you need quick, private web searches in AI workflows, such as fetching real-time data for responses, avoiding data collection in user queries, or integrating search into tools that require filtered results (e.g., safe search for family-friendly content).

Key Capabilities

• Perform searches with parameters like query string, result count (e.g., up to 50), and filters for safe search, language (e.g., "en-US"), and country (e.g., "US").

• Extract snippets: API returns JSON with title, URL, and description fields for each result.

• Privacy features: No user tracking; results based on Brave's index.

• Rate limiting: Up to 1000 requests per day per API key, depending on plan.

Usage Patterns

Always authenticate requests using the $BRAVE_API_KEY environment variable. Make HTTP GET requests to the endpoint with required parameters. Structure queries as URL-encoded strings. For AI agents, integrate via HTTP clients in languages like Python or curl. Test in a loop for pagination by incrementing the 'offset' parameter (e.g., start at 0, then 10).

Common Commands/API

Endpoint: https://api.search.brave.com/res/v1/web/search

Authentication: Include header 'X-Subscription-Token: $BRAVE_API_KEY' (set via export BRAVE_API_KEY=your_key in your environment).

Key parameters: q (query string), count (number of results, e.g., 10), offset (for pagination, e.g., 0), search_lang (e.g., "en"), country (e.g., "US"), safe (e.g., "active" for safe search).

Example curl command:

curl -H "X-Subscription-Token: $BRAVE_API_KEY" "https://api.search.brave.com/res/v1/web/search?q=AI+assistants&count=5&search_lang=en"

Python snippet using requests:

import requests; import os;
response = requests.get('https://api.search.brave.com/res/v1/web/search', headers={'X-Subscription-Token': os.environ['BRAVE_API_KEY']], params={'q': 'open source AI', 'count': 3})

Config format: Store API key in .env files as BRAVE_API_KEY=your_key, and load with a library like python-dotenv.

Integration Notes

Integrate by wrapping API calls in your AI agent's code; use async requests for non-blocking operations. For example, in OpenClaw, call this skill before generating responses that require external data. Handle config by exporting $BRAVE_API_KEY or reading from a secure vault. If combining with other AI APIs, chain this skill after tools like language models to enrich outputs (e.g., search for context, then summarize). Avoid caching results longer than 1 hour due to web content volatility.

Error Handling

Check HTTP status codes: 200 for success, 401 for invalid API key (retry after verifying $BRAVE_API_KEY), 429 for rate limit (wait and retry with exponential backoff, e.g., 5-10 seconds). Parse JSON responses for errors like "error": "Bad request" and log details. In code, use try-except blocks:

try: response = requests.get(...)
except requests.exceptions.RequestException as e: print(f"Error: {e}"); exit(1)

For CLI, pipe output to grep for error patterns and handle failures by exiting with a non-zero code.

Concrete Usage Examples

• Search for latest AI news: Use to fetch headlines on AI topics. Command: curl -H "X-Subscription-Token: $BRAVE_API_KEY" "https://api.search.brave.com/res/v1/web/search?q=latest+AI+advancements&count=3&search_lang=en". In your agent, extract the first result's snippet and append to a response.

• Extract snippets for query analysis: For a user query like "best open source tools", run: curl -H "X-Subscription-Token: $BRAVE_API_KEY" "https://api.search.brave.com/res/v1/web/search?q=best+open+source+tools&safe=active&country=US". Process the JSON to pull descriptions, then use in AI reasoning to recommend tools based on extracted data.

Graph Relationships

• Related to: ai-apis (cluster), as it provides search capabilities for other AI tools.

• Connects with: Other search APIs in ai-apis for fallback options, or data processing skills for result analysis.