| name | exa-search |
| description | Load when the user needs web search, code examples, company intel, people lookup, or current information. Use for queries like 'search for', 'look up', 'find', 'what's the latest on', 'research company', 'code examples for'. Does not require a browser. |
Exa Search
Neural search via Exa MCP for web content, code, companies, and people.
Setup
Requires EXA_API_KEY environment variable (get one at exa.ai).
Config is in mcporter.json at the skill directory. Verify with:
npx mcporter list exa --config "${SKILL_DIR}/mcporter.json" --schema
Tools
All commands use --config "${SKILL_DIR}/mcporter.json" --output json.
web_search_exa
General web search. Use for most queries.
npx mcporter call exa.web_search_exa query="natural language query" numResults:5 --config "${SKILL_DIR}/mcporter.json" --output json
Query style: Describe the ideal page, not keywords. "blog post comparing React and Vue performance" not "React vs Vue".
Use category:people or category:company prefix for LinkedIn/company searches.
web_search_advanced_exa
Use when you need filters: domains, dates, categories, or content options.
Key filters:
includeDomains / excludeDomains — domain filtering
startPublishedDate / endPublishedDate — ISO date (YYYY-MM-DD)
category — "company", "research paper", "news", "pdf", "github", "people", etc.
enableHighlights / enableSummary — richer content extraction
get_code_context_exa
Use for code examples, API docs, and programming solutions.
npx mcporter call exa.get_code_context_exa query="React useState hook examples" numResults:5 --config "${SKILL_DIR}/mcporter.json" --output json
Query style: Be specific. "Python requests library POST with JSON body" not "python http".
Tool Selection
| Need | Tool |
|---|
| Quick answer | web_search_exa with numResults:3-5 |
| Code/API docs | get_code_context_exa |
| Domain/date filters | web_search_advanced_exa |
| Company research | web_search_advanced_exa with category:"company" |
| People search | web_search_advanced_exa with category:"people" |
| Research papers | web_search_advanced_exa with category:"research paper" |
Gotchas
EXA_API_KEY is required — the server returns 401 without it
- Use natural language queries, not keyword-style — Exa is semantic search
numResults max is 100 for web_search_exa, 20 for get_code_context_exa
includeDomains takes a JSON array string: '["github.com","react.dev"]'
- Results include highlights by default; use
enableSummary for AI summaries