# Exa AI MCP

> Exa AI MCP connects your agent to advanced web intelligence for deep research and analysis. It moves beyond basic keyword searches, allowing you to query the web based on meaning and context. You can instantly find sites similar to a target URL, pull clean text from any page, or search and retrieve content all in one go. Exa lets your AI client act like a real-time analyst for deep market research or competitor tracking.

## Overview
- **Category:** artificial-intelligence
- **Price:** Free
- **Tags:** semantic-search, web-discovery, information-retrieval, content-analysis, url-matching

## Description

Stop bouncing between browser tabs just to gather context. This MCP gives your agent the power of semantic web discovery. You can tell it what you need—like understanding the trends around 'climate change mitigation'—and it searches the web based on meaning, not just matching keywords. Need to know who else is covering that topic? Your agent finds similar websites instantly. It also pulls clean text from complex pages so your workflow doesn't get bogged down in HTML junk. If you’re doing deep market research or auditing sources for a project, this MCP ensures your answers are grounded in real-time, relevant data. Because Vinkius hosts this MCP, you connect once to your preferred AI client and gain access to sophisticated web analysis that turns complex information gathering into simple conversation.

## Tools

### find_similar
Finds websites that are similar to a given URL; note this endpoint is deprecated and should not be used.

### get_contents
Retrieves the cleaned text content specifically for provided page IDs.

### search_with_contents
Performs a search across the web and pulls both the results and their full page contents in one request.

### search_web
Runs a semantic query against the open web, returning relevant URLs, titles, and metadata.

## Prompt Examples

**Prompt:** 
```
Find 5 high-quality research papers about 'Climate Change mitigation' using Exa.
```

**Response:** 
```
I've retrieved 5 highly relevant research papers. Notable results include studies on carbon capture and renewable energy adoption. Would you like the links and summaries for each?
```

**Prompt:** 
```
Find sites similar to https://techcrunch.com.
```

**Response:** 
```
I've found 10 sites similar to TechCrunch. Top matches include VentureBeat, The Verge, and Wired. I can provide the relevancy scores for each if you'd like.
```

**Prompt:** 
```
Search for 'future of AI robotics' and give me the full content of the best result.
```

**Response:** 
```
Search complete! The top result is a detailed analysis from MIT. I've retrieved the cleaned content, which discusses embodied AI and humanoid developments. Would you like a summary?
```

## Capabilities

### Perform semantic searches
Query the open web using natural language, retrieving results based on meaning rather than literal keywords.

### Identify related websites
Find websites that share characteristics with a given URL for competitor analysis or source mapping.

### Extract structured content
Retrieve clean, usable text and HTML content from specific web pages without manual scraping.

### Execute combined search tasks
Run a single request that searches the web and pulls the full page contents simultaneously.

### Monitor API usage
Track your research budget by monitoring API calls and crawl statistics within your agent's workflow.

## Use Cases

### Competitor Landscape Audit
A marketing analyst needs to know who else is covering a new industry standard. Instead of manually Googling, they prompt their agent to run `search_web` and then use the similar discovery capability to build a comprehensive list of related publications.

### Academic Literature Review
A PhD candidate needs deep context on 'renewable energy policy'. They ask their agent to perform a semantic search, retrieve multiple sources using `search_with_contents`, and analyze the full text against their hypothesis in one step.

### Technical Documentation Check
A developer needs to verify a specific piece of documentation online. They use the MCP's content retrieval tools to pull clean, verifiable text from a URL ID, ensuring their agent is working with accurate data, not just snippets.

### Rapid Trend Monitoring
An operations lead needs to monitor niche industry trends daily. They set up an automated workflow that runs `search_web` on key phrases and flags any major changes in related websites or coverage.

## Benefits

- Stop wading through messy HTML. Use `get_contents` to pull clean, structured text from any web page directly into your agent's memory.
- Identify competitors instantly. Run a check for similar sites using the appropriate tool to map out related sources or potential market overlaps.
- Cut down on requests by combining actions. The `search_with_contents` tool lets you search and retrieve full content in one go, speeding up complex workflows.
- Keep your research budget tight. Monitor API usage directly through the MCP to maintain strict control over your web discovery spending.
- Get deeper results with semantic power. Use `search_web` for queries that rely on meaning rather than just matching keywords.

## How It Works

The bottom line is that you get continuous web analysis piped directly into your conversational workflow.

1. Subscribe to this MCP on Vinkius and provide your Exa API Key.
2. Connect the MCP to your preferred AI client (like Claude or Cursor).
3. Prompt your agent with a request, such as 'Search for X and give me the content of the top result.' The agent executes the search, pulls the data, and presents it.

## Frequently Asked Questions

**How does Exa AI MCP perform semantic searching?**
Exa AI performs semantic searches using meaning-based embeddings rather than simple keyword matching. This means your agent understands the intent of your query and finds results based on topic relevance, even if keywords aren't used.

**Can I use Exa AI MCP to monitor my usage?**
Yes, you can track your research budget. The MCP provides monitoring tools so your agent workflow keeps a count of API calls and crawl statistics, keeping you in control of costs.

**What is the difference between `search_web` and `search_with_contents`?**
`search_web` returns metadata—just the URLs, titles, and summaries. However, `search_with_contents` performs both actions in one request, giving you the full, clean text content of the best results instantly.

**Does Exa AI MCP help with competitor analysis?**
Absolutely. You can run a query to find sites similar to a known URL, helping you map out competitors or related sources for market research purposes.

**Is the `find_similar` tool still good to use in Exa AI MCP?**
No. The listing notes that `find_similar` is deprecated by Exa. You should rely on the semantic search capabilities of the other tools for accurate site similarity checks.