AgentSkillsCN

exa-search

借助Exa API进行网页搜索、代码上下文查询、公司研究与深度调研。在进行网络搜索、查找代码示例与文档、研究公司、爬取URL，或执行深度AI研究任务时使用。可触发的场景包括：网页搜索、查找代码、研究公司、爬取页面、人物搜索。

SKILL.md
--- frontmatter
name: exa-search
description: "Web search, code context, company research, and deep research via Exa API. Use when searching the web, finding code examples/docs, researching companies, crawling URLs, or running deep AI research tasks. Triggers on: web search, find code, research company, crawl page, people search."

Exa Search

Search the web and extract content via Exa's AI-powered search API.

Installation

bash
mcporter install exa-cli

CLI: exa-cli

All commands support --output json for machine-readable output.

Web Search

bash
# Basic search
exa-cli web-search-exa --query "latest AI news"

# With options
exa-cli web-search-exa --query "React hooks tutorial" --num-results 5 --type fast

Code Context

Find code examples, docs, and solutions from GitHub, Stack Overflow, and official docs.

bash
exa-cli get-code-context-exa --query "Node.js streams examples"
exa-cli get-code-context-exa --query "Python pandas filtering" --tokens-num 10000

Company Research

bash
exa-cli company-research-exa --company-name "Anthropic"
exa-cli company-research-exa --company-name "OpenAI" --num-results 5

URL Crawling

Extract full content from a specific URL.

bash
exa-cli crawling-exa --url "https://example.com/article"
exa-cli crawling-exa --url "https://docs.example.com" --max-characters 5000

People Search

Find professionals and their profiles.

bash
exa-cli people-search-exa --query "AI researchers at Google"

Advanced Search

Full control over filters, domains, dates, and content options.

bash
exa-cli web-search-advanced-exa --query "machine learning" \
  --category "research paper" \
  --include-domains "arxiv.org,github.com" \
  --num-results 10

Categories: company, research paper, news, pdf, github, tweet, personal site, people, financial report

Deep Research

Start an AI research agent for complex questions (15s-2min).

bash
# Start research
exa-cli deep-researcher-start --instructions "Compare transformer architectures for NLP"

# Check status (use returned task ID)
exa-cli deep-researcher-check --task-id "abc123"

Models: exa-research (faster, default) | exa-research-pro (comprehensive)

Notes

  • •No API key needed (uses hosted endpoint at mcp.exa.ai)
  • •Returns clean, LLM-ready text content
  • •Deep research runs async; poll with deep-researcher-check