AgentSkillsCN

exa-search

使用 exa CLI 通过 Exa 的 API 搜索网络。当用户要求运行 Exa 搜索、获取内容、查找相似链接或使用 exa 命令行工具获取答案时，请使用此功能。

SKILL.md
--- frontmatter
name: exa-search
description: >
  Use the exa CLI to search the web via Exa's API. Use when the user asks to
  run Exa search, fetch contents, find similar links, or get answers using the
  exa command-line tool.

Exa Search CLI

Agent-first, non-interactive CLI wrapper for Exa Search API. Default output is JSON.

Quick start

  1. •Ensure EXA_API_KEY is set.
  2. •Run exa --help to see all commands and examples.
  3. •Use exa search, exa contents, exa find-similar, or exa answer.
  4. •Use --plain for stable, line-based output when needed.

Commands

  • •exa search <query>: semantic search; supports filters and content options.
  • •exa contents <url...>: fetch contents for URLs.
  • •exa find-similar <url>: find similar links to a URL.
  • •exa answer <query>: grounded Q&A; supports --stream.

Output modes

  • •Default: JSON to stdout.
  • •--plain: stable, line-based output.
  • •Errors: stderr (JSON unless --plain).

Shared content options

Available on search, contents, and find-similar:

  • •--text, --text-max-characters, --text-include-html-tags
  • •--highlights, --highlights-num-sentences, --highlights-per-url, --highlights-query
  • •--summary, --summary-query, --summary-schema <json|@file>
  • •--context, --context-max-characters
  • •--livecrawl <never|fallback|preferred|always>, --livecrawl-timeout <ms>
  • •--subpages, --subpage-target
  • •--extras-links, --extras-image-links

Examples

Search:

bash
exa search "latest developments in quantum computing"
exa search "AI chips roadmap" --type deep --additional-query "GPU roadmap" --num-results 25
exa search "LLM hallucinations" --include-domain arxiv.org --text --text-max-characters 2000

Find similar:

bash
exa find-similar https://arxiv.org/abs/2307.06435 --num-results 5 --text
exa find-similar https://example.com --exclude-domain example.com --summary

Contents:

bash
exa contents https://arxiv.org/abs/2307.06435 --text --summary --summary-query "key findings"
exa contents https://example.com --highlights --highlights-per-url 2 --highlights-num-sentences 2

Answer:

bash
exa answer "What is the population of New York City?"
exa answer "state of solid-state batteries" --stream

Plain output:

bash
exa search "openai" --num-results 3 --plain
exa find-similar https://openai.com --num-results 3 --plain
exa contents https://openai.com --plain
exa answer "What is the capital of France?" --plain

Dry run:

bash
exa search "openai" --num-results 1 --dry-run
exa contents https://openai.com --text --dry-run