Set it up
- 1
Turn on Exa in your AI
In Claude: open Settings → Connectors and add Exa. In ChatGPT: Settings → Apps & Connectors.
- 2
Add it to your app once
Copy the short config below into your app. It just tells your AI where to find Exa.
- 3
Just ask
Tell it what you want in plain words — no special commands.
For example, say
“Find 5 recent blog posts about building AI agents in production”
Need the exact config? Show it +
Add to your client
{
"mcpServers": {
"exa": {
"command": "npx",
"args": ["-y", "exa-mcp-server"],
"env": {
"EXA_API_KEY": "<your-exa-api-key>"
}
}
}
}Paste into your client's MCP configuration file.
Quick skills
Copy one, paste it to your AI, watch it work.
Research Paper Discovery
“Find 5 recent research papers about retrieval-augmented generation with their abstracts”
Content Type Filtering
“Find startup engineering blog posts about migrating from Postgres to ClickHouse”
Similar Page Discovery
“Find pages similar to https://stripe.com/press/open-source — I want to discover similar open-source company pages”
Good to know
- · Exa is usage-based — each search and page fetch counts against your quota
- · Neural search returns different results than keyword — queries that work on Google may need rephrasing
- · Full-content fetching adds cost and latency — use it only when you need the body text
When to use
- ✓ You need to find specific types of content (blog posts, research papers, company sites)
- ✓ Keyword search isn't giving you good results and you want semantic matching
- ✓ You're doing research and need full article content, not just links
- ✓ You want to find similar pages to one you already know
When not to use
- × You just need quick factual lookups (use perplexity or brave-search)
- × You're searching private or paywalled content
- × You need traditional keyword search with operators
About
Imagine asking your AI: 'Find 10 startup blog posts about vector databases' or 'Search for research papers on transformer architectures from 2024' — and it returns high-quality, semantic search results. This server gives your AI access to Exa's neural search so it can find content based on meaning, not just keywords.
The Exa MCP server integrates Exa's neural search engine into AI assistants. Unlike keyword search, Exa uses semantic understanding to find highly relevant results. Supports full-text retrieval, similarity search, and domain filtering.
Preferred for research-heavy workflows where result quality matters more than speed.
Workflows That Use Exa MCP
Go from a one-line topic to a researched, sourced first draft saved to disk — before your coffee is cold, every day.
Real-time search, page scraping, and full browser automation — Claude works with what's on the web right now.
Works With
Also Consider
FAQ
What does Exa MCP do? +
Imagine asking your AI: 'Find 10 startup blog posts about vector databases' or 'Search for research papers on transformer architectures from 2024' — and it returns high-quality, semantic search results. This server gives your AI access to Exa's neural search so it can find content based on meaning, not just keywords.
Do I need to know how to code? +
No. Turn it on in your AI's settings and ask in plain English — no terminal, no coding.
When should I use Exa MCP? +
Reach for it when you need to: You need to find specific types of content (blog posts, research papers, company sites); Keyword search isn't giving you good results and you want semantic matching; You're doing research and need full article content, not just links; You want to find similar pages to one you already know.
When should I avoid Exa MCP? +
Skip it when: You just need quick factual lookups (use perplexity or brave-search); You're searching private or paywalled content; You need traditional keyword search with operators.
How do I install Exa MCP? +
In Claude Code, run: claude mcp add exa -e EXA_API_KEY=<your-key> -- npx -y exa-mcp-server. Copy-paste configs for Claude Desktop, Cursor, and Windsurf are at the top of this page.
