exa-search

advait's avatarfrom advait

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.

⭐0stars🔀0forks📁View on GitHub🕐Updated Jan 2, 2026
[Web Research]

When & Why to Use This Skill

This Claude skill leverages the Exa Search API to provide advanced semantic web search and content retrieval capabilities. It allows Claude to perform deep research, find similar links, and extract clean, structured content from the web, enabling the agent to provide grounded, real-time answers to complex queries.

Use Cases

  • Technical Deep-Dives: Conduct comprehensive research on specialized topics like 'AI chip roadmaps' or 'quantum computing' using semantic search and domain filtering.
  • Automated Content Extraction: Fetch and summarize the main findings from specific URLs or academic papers without manual browsing.
  • Discovery and Benchmarking: Use the 'find-similar' feature to identify related websites, competitors, or research papers based on a provided URL.
  • Grounded Fact-Checking: Provide accurate answers to factual questions by searching the live web and retrieving highlights from authoritative sources.
nameexa-search
description>

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:

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:

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:

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:

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

Plain output:

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:

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