exa-search

name	exa-search
description	Web search and content extraction via Exa public MCP endpoint. Use for searching documentation, facts, or any web content. No API key required. Lightweight, no browser required.

Exa Search

Web search and content extraction using Exa's public MCP endpoint. No API key required.

Setup

No setup needed — no API key, no dependencies (pure fetch).

Search

{baseDir}/search.js "query"                           # Basic search (5 results)
{baseDir}/search.js "query" -n 10                     # More results
{baseDir}/search.js "query" --highlights              # Include relevant highlights
{baseDir}/search.js "query" --text                    # Include full page text
{baseDir}/search.js "query" --highlights --text       # Both highlights and text
{baseDir}/search.js "query" --type fast               # Fast search for quick results
{baseDir}/search.js "query" --start-date 2024-01-01   # Filter by published date
{baseDir}/search.js "query" --domains "docs.rs,github.com"  # Limit to specific domains
{baseDir}/search.js "query" -n 3 --highlights         # Focused search with highlights

Options

• -n <num> - Number of results (default: 5)
• --highlights - Include relevant highlights from each page
• --text - Include full text content from each page
• --type <type> - Search type: auto (default), fast
• --start-date <date> - Start published date (YYYY-MM-DD)
• --end-date <date> - End published date (YYYY-MM-DD)
• --domains <list> - Comma-separated domains to include
• --exclude-domains <list> - Comma-separated domains to exclude

Extract Page Content

{baseDir}/content.js https://example.com/article                          # Extract text
{baseDir}/content.js https://example.com/article --highlights             # Auto highlights
{baseDir}/content.js https://example.com/article --highlights "AI"        # Query-guided highlights
{baseDir}/content.js url1 url2 url3                                       # Multiple URLs at once

Options

• --highlights [query] - Include highlights, optionally guided by a search query
• --max-chars <n> - Max characters for text (default: 5000)
• --text - Full text (default behavior)

Output Format

The MCP endpoint returns formatted text content directly, including titles, URLs, dates, and page content.

When to Use

• Searching for documentation or API references
• Looking up facts or current information
• Fetching content from specific URLs
• Any task requiring web search without interactive browsing
• When you need highlights (more token-efficient than full text)
• When you need domain-filtered or date-filtered results