Exa MCP Server
✓ Text source-checked Oct 27, 2026Exa MCP for web search and web crawling
Source-checked against the vendor's site · rebuilt from the vendor's own pagesProfile updated Link checked Pricing captured How we rate
In short
Exa MCP Server connects AI agents to Exa for web search, content fetching, and multi-step research via a hosted endpoint.
Exa MCP Server connects AI agents to Exa for web search, content fetching, and multi-step research. The service is available as a hosted MCP server at https://mcp.exa.ai/mcp.
The default tool web_search_exa searches the web for any topic and returns clean, ready-to-use content. The default tool web_fetch_exa reads a webpage's full content as clean markdown from one or more URLs. The optional tool agent_run runs an Exa Agent for multi-step research, list-building, enrichment, and structured output. The optional tool web_search_advanced_exa provides advanced search with filters, domains, dates, highlights, summaries, and subpage crawling.
The hosted MCP server works anonymously with rate limits. Users can use OAuth or an API key for higher limits and access to the Exa Agent.
Exa MCP works with most other clients by pointing them at the hosted endpoint. Agent skills load with Agent Plugin or Claude plugin installs.
According to Exa MCP Server: Exa MCP Server connects AI agents to Exa for web search, Content fetching, and multi-step research. Its pages list this strength: Exa MCP Server is available as a hosted endpoint, removing the need for local installation. They also state this limitation: Anonymous usage of the hosted server is subject to rate limits.
— The AI Tool Bible editorial team
Pros
- ✅ Exa MCP Server is available as a hosted endpoint, removing the need for local installation.
- ✅ The web search tool returns clean, ready-to-use content for any topic.
- ✅ The web fetch tool provides full webpage content as clean markdown.
- ✅ The agent tool supports multi-step research, list-building, enrichment, and structured output.
- ✅ The hosted server allows anonymous usage with rate limits.
Cons
- ⚠️ Anonymous usage of the hosted server is subject to rate limits.
- ⚠️ The Exa Agent tool requires authentication via OAuth or an API key.
Pricing (captured )
| Tier | Monthly | Per seat | Key features |
|---|---|---|---|
| Free | $0 USD per month | — |
|
| Team | $4 USD per user/month | Yes |
|
| Enterprise | $21 USD per user/month | Yes |
|
Prices change; we show the grid as captured on Oct 27, 2026. Check current pricing on the vendor's site →
Facts
- Pricing
- freemium - Free tier · from $4 USD/mo (not source-checked)
- API available
- Yes
- Free trial
- Yes (not source-checked)
- Open source
- Yes (license: MIT) (not source-checked)
- GitHub stars
- 5,199 (checked Oct 29, 2026) (not source-checked)
- Last GitHub push
- Oct 28, 2026 (not source-checked)
- First commit
- Nov 2024 (not source-checked)
- Model used
- Exa neural search index (in-house) (not source-checked)
Where this tool sits
Use cases
Frequently asked
- How much does Exa MCP Server cost?
- Exa MCP Server uses a freemium model. There is a free tier available, and paid plans start from $4 USD per month. Users can also use OAuth or an API key for higher limits.
- What tools are included in the Exa MCP Server?
- It includes web_search_exa for web search, web_fetch_exa for reading webpage content as markdown, agent_run for multi-step research, and web_search_advanced_exa for filtered searches with highlights and summaries.
- Do I need an API key to use Exa MCP Server?
- No, the hosted MCP server works anonymously with rate limits. However, using OAuth or an API key provides higher limits and access to the Exa Agent for multi-step research tasks.
- Which AI clients are compatible with Exa MCP Server?
- Exa MCP works with most other clients by pointing them at the hosted endpoint. Agent skills can also be loaded using Agent Plugin or Claude plugin installs.
- What is the primary use case for Exa MCP Server?
- It is designed for web search for AI agents, content fetching, multi-step research, list-building, and advanced search with filters. It returns clean, ready-to-use content for AI applications.