# Exa MCP Tools Reference

## web_search_exa

Basic web search with clean, ready-to-use content.

**Parameters:**
- `query` (string, required): Search query
- `numResults` (number, optional): Number of results (default: 10)

**Example:**
```bash
exa-cli.mjs search "AI agent frameworks 2026"
```

---

## web_search_advanced_exa

Advanced search with full control over filters.

**Parameters:**
- `query` (string, required): Search query
- `numResults` (number, optional): Number of results
- `type` (string, optional): "auto", "neural", "deep"
- `category` (string, optional): "company", "news", "tweet", "people", "personal site"
- `includeDomains` (array, optional): Only include these domains
- `excludeDomains` (array, optional): Exclude these domains
- `startPublishedDate` (string, optional): ISO date YYYY-MM-DD
- `endPublishedDate` (string, optional): ISO date YYYY-MM-DD
- `livecrawl` (string, optional): "never", "fallback", "always"

**Category Restrictions:**
- `category: "company"`: Cannot use domain or date filters
- `category: "people"`: Cannot use date filters, excludeDomains, or excludeText
- `includeText`/`excludeText`: Only single-item arrays

**Examples:**
```bash
# Domain filter
exa-cli.mjs search-advanced "AI news" --domains techcrunch.com,wired.com

# Date filter
exa-cli.mjs search-advanced "OpenAI" --after 2026-01-01 --before 2026-02-01

# More results
exa-cli.mjs search-advanced "machine learning" --num 30
```

---

## get_code_context_exa

Find code examples, documentation, and solutions from GitHub, StackOverflow, and technical docs.

**Parameters:**
- `query` (string, required): Code-related query
- `tokensNum` (number, optional): Token limit (default: 5000, range: 1000-50000)

**Best Practices:**
- Always include programming language in query
- Include framework + version when relevant
- Use exact identifiers (function names, error messages)

**Token Strategy:**
- Focused snippet: 1000-3000
- Most tasks: 5000
- Complex integration: 10000-20000

**Examples:**
```bash
# Python specific
exa-cli.mjs code "Python asyncio retry exponential backoff"

# Framework specific
exa-cli.mjs code "Next.js 14 server actions form handling"

# Error lookup
exa-cli.mjs code "ECONNREFUSED Node.js Docker"
```

---

## crawling_exa

Get full content from a specific URL.

**Parameters:**
- `url` (string, required): URL to crawl

**Example:**
```bash
exa-cli.mjs crawl "https://docs.anthropic.com/claude/docs"
```

---

## company_research_exa

Research any company for business information, news, and insights.

**Parameters:**
- `companyName` (string, required): Company name to research

**Returns:**
- Company overview
- Recent news
- Key metrics (when available)
- Competitors
- Funding info

**Example:**
```bash
exa-cli.mjs company "Stripe"
exa-cli.mjs company "Klarna fintech"
```

---

## people_search_exa

Find people and their professional profiles.

**Parameters:**
- `query` (string, required): Search query (role, name, company, etc.)
- `numResults` (number, optional): Number of results

**Notes:**
- Returns public LinkedIn profiles and professional bios
- Works best with role + location/company

**Examples:**
```bash
exa-cli.mjs people "VP Engineering AI startups San Francisco"
exa-cli.mjs people "CTO fintech Stockholm"
```

---

## deep_researcher_start

Start an AI research agent that searches, reads, and writes a detailed report.

**Parameters:**
- `query` (string, required): Research topic

**Returns:**
- `taskId`: Use with `research-check` to get results

**Note:** Deep research is async. Poll with `research-check` until status is "completed".

**Example:**
```bash
exa-cli.mjs research "Current state of nuclear fusion energy startups 2026"
# Returns: taskId: abc123...

exa-cli.mjs research-check abc123
# Poll until complete
```

---

## deep_researcher_check

Check status and get results from a deep research task.

**Parameters:**
- `taskId` (string, required): Task ID from `deep_researcher_start`

**Status Values:**
- `pending`: Still processing
- `completed`: Results ready
- `failed`: Error occurred

**Example:**
```bash
exa-cli.mjs research-check abc123-def456-...
```

---

## Rate Limits

**Free Plan:**
- ~1 request/second
- 2000 requests/month

**With API Key:**
- Higher limits based on plan
- Get key: https://dashboard.exa.ai/api-keys

Set key:
```bash
export EXA_API_KEY="your-key-here"
```

---

## MCP Server URL

Base: `https://mcp.exa.ai/mcp`

With all tools:
```
https://mcp.exa.ai/mcp?tools=web_search_exa,web_search_advanced_exa,get_code_context_exa,crawling_exa,company_research_exa,people_search_exa,deep_researcher_start,deep_researcher_check
```

With API key:
```
https://mcp.exa.ai/mcp?exaApiKey=YOUR_KEY&tools=...
```