Tool
Call the native MCP tool exa-search__web_search_advanced_exa directly. Parameters go in a JSON object.
Critical Rules
- Never run searches in main context. Spawn subagents or run sequentially.
- Query variation: 2-3 phrasings, run in parallel, deduplicate.
- Dynamic numResults: "a few" → 10-20, "comprehensive" → 50-100.
Categories
| Category | Use for |
|---|---|
people |
LinkedIn profiles, public bios (primary) |
personal site |
Blogs, portfolios, about pages |
news |
Press mentions, interviews, speaker bios |
| (none) | General web, broader context |
Restrictions with category=people
Do NOT use: startPublishedDate, endPublishedDate, startCrawlDate, endCrawlDate, excludeDomains, includeText, excludeText.
includeDomains → LinkedIn only: ["linkedin.com"].
Call Examples
Tool: exa-search__web_search_advanced_exa
# Find people by role
{ query: "VP Engineering AI infrastructure", category: "people", numResults: 20, type: "auto" }
# Deep dive on specific person
{ query: "Dario Amodei Anthropic CEO background", type: "auto", livecrawl: "fallback", numResults: 15 }
# News mentions
{ query: "Dario Amodei interview 2024", category: "news", numResults: 10, startPublishedDate: "2024-01-01" }
# LinkedIn only
{ query: "Machine learning engineer San Francisco", category: "people",
includeDomains: ["linkedin.com"], numResults: 25 }
Output Format
Return:
- Results — name, title, company, location (if available)
- Sources — profile URLs
- Notes — profile completeness, verification status
Browser Fallback
If content is auth-gated (private LinkedIn, etc.) → use browser tool with logged-in session.