AgentSkillsCN

exa

通过 MCP 利用 Exa AI 在网络中搜索、调研公司，并查找代码上下文。当您被要求搜索信息、研究主题、查阅文档、寻找代码示例、调查公司，或爬取特定 URL 时，可使用此技能。当您发出类似“搜索……”、“查找……相关信息”、“研究……”、“查阅……”、“……是什么？”、“公司调研”等查询时，该技能便会自动触发。

SKILL.md
--- frontmatter
name: exa
description: Search the web, research companies, and find code context using Exa AI via MCP. Use when asked to search for information, research topics, look up documentation, find code examples, investigate companies, or crawl specific URLs. Triggers on queries like "search for...", "find info about...", "research...", "look up...", "what is...", "company research", or any web lookup task.
metadata: {"clawdbot":{"emoji":"🔍","requires":{"bins":["mcporter"]}}}

Exa Search

Use mcporter to call Exa MCP tools for web search, research, and code context.

Quick Reference

bash
# Web search
mcporter call exa.web_search_exa query="your search" --config ~/clawd/config/mcporter.json

# Deep research (comprehensive)
mcporter call exa.deep_search_exa objective="what you want to find" --config ~/clawd/config/mcporter.json

# Code/API docs
mcporter call exa.get_code_context_exa query="React hooks tutorial" --config ~/clawd/config/mcporter.json

# Crawl specific URL
mcporter call exa.crawling_exa url="https://example.com" --config ~/clawd/config/mcporter.json

# Company research
mcporter call exa.company_research_exa companyName="Stripe" --config ~/clawd/config/mcporter.json

Tools

ToolUse When
web_search_exaGeneral web search, news, articles
deep_search_exaComprehensive research on complex topics
get_code_context_exaFinding code examples, API docs, library usage
crawling_exaExtracting content from a known URL
company_research_exaBusiness intelligence, company info

Parameters

web_search_exa

  • •query (required): Search query
  • •numResults: Number of results (default: 8)
  • •type: auto | fast | deep

deep_search_exa

  • •objective (required): Natural language description of what to find
  • •search_queries: Optional keyword list

get_code_context_exa

  • •query (required): Code/API search query
  • •tokensNum: Token limit 1000-50000 (default: 5000)

crawling_exa

  • •url (required): URL to crawl
  • •maxCharacters: Max chars (default: 3000)

company_research_exa

  • •companyName (required): Company to research
  • •numResults: Number of results (default: 5)

Tips

  • •Add --output json for structured output
  • •Use deep_search_exa for nuanced research questions
  • •Use get_code_context_exa for any programming question — it searches GitHub and StackOverflow
  • •Prefer web_search_exa with type=fast for quick lookups