Exa Web Search & Content Extraction

Real-time web search and content extraction powered by Exa AI.

Usage

Use the Python script at scripts/exa.py. Requires EXA_API_KEY environment variable.

Web Search

# Basic search
python scripts/exa.py search --query "latest AI research"

# Fast search for simple queries
python scripts/exa.py search --query "node.js version" --type fast

# Deep search for comprehensive results
python scripts/exa.py search --query "React server components" --type deep --num-results 10

# With full text content
python scripts/exa.py search --query "GraphQL best practices" --text

# Filter by domain
python scripts/exa.py search --query "LLM research" --include-domains "arxiv.org,paperswithcode.com"

# Filter by date
python scripts/exa.py search --query "AI news" --start-date "2025-01-01T00:00:00.000Z"

Extract URL Content

# Extract from documentation
python scripts/exa.py contents --urls "https://docs.python.org/3/tutorial/classes.html"

# Multiple URLs
python scripts/exa.py contents --urls "https://example.com/page1,https://example.com/page2"

# Prefer live content
python scripts/exa.py contents --urls "https://example.com/page" --livecrawl preferred

Code Examples

# Find code examples
python scripts/exa.py code --query "React useState hook examples"

# Quick syntax lookup
python scripts/exa.py code --query "Python list comprehension" --tokens 2000

# Detailed patterns
python scripts/exa.py code --query "GraphQL resolver patterns" --tokens 15000

Search Types

Type	Use Case	Speed
fast	Quick lookups, simple queries	Fastest
auto	General purpose (default)	Medium
deep	Complex research, comprehensive	Slowest

When to Use Each Command

Use search	Use contents	Use code
Finding relevant pages	Have a specific URL	Programming questions
General web research	Extracting known content	API/library usage
News and articles	Reading documentation	Code examples

Query Tips

• Be specific: "React useEffect cleanup function examples" not "useEffect"
• Include the library/framework name
• Specify the programming language
• Mention the use case (authentication, caching, etc.)

Rules

• Use --type fast for simple factual queries
• Use --type deep for research requiring comprehensive results
• Use --text to get full page content in search results
• Use code command for programming questions instead of search
• Use contents when you have a specific URL to extract
• Use --format json for structured output when needed