Guide · Agent skill
How to use the Exa Neural Search agent skill (exa-search)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The exa-search skill gives an AI agent a set of tools for searching the web through Exa, a neural search service, instead of relying only on what the model already knows. This guide explains what the skill does, how to install it, how to ask for it in plain language, and what it needs to run. By the end you will be able to set up the Exa MCP server, trigger a search, and know when the skill is worth its cost.
Exa Neural Searchaffaan-m/ECC · MIT Download the skill Quick start
- 1Get an Exa API key. Create an account at exa.ai and copy your API key.
- 2Install Node.js and npm. The MCP server runs through npx, so Node.js with npm must be available on your machine.
- 3Download and unzip the skill. Download the exa-search ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
- 4Add the MCP server to ~/.claude.json. Add the exa-web-search entry with your key in the EXA_API_KEY field.
- 5Ask for a search in plain language. Request something such as looking up recent news or finding code examples, and the agent will choose the matching tool.
- 6Check results and refine. Narrow with domain or date filters, or use crawling_exa to read a specific page in full.
What the exa-search skill lets the agent do
Once installed, the agent can run searches through Exa instead of answering only from its built-in knowledge. The skill covers current news and web pages, code examples and API documentation, company background, people profiles, and full-page reading from a URL.
It also supports a longer background research job: the agent starts the job, does other work, and checks back for results.
- General web search for current information, news, or facts (web_search_exa)
- Filtered web search by domain and publication date (web_search_advanced_exa)
- Code examples and documentation from GitHub, Stack Overflow, and docs sites (get_code_context_exa)
- Company research for business intelligence and news (company_research_exa)
- Professional profiles and bios (people_search_exa)
- Full page content extraction from a URL (crawling_exa)
- Asynchronous deep research (deep_researcher_start and deep_researcher_check)
Install the exa-search skill
Download the skill ZIP from this page and unzip it. For a personal install, place the folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
The package contains two files: SKILL.md, which holds the instructions the agent reads, and agents/openai.yaml.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills
Configure the Exa MCP server and API key
The skill does not work until the Exa MCP server is configured. The server is the npm package exa-mcp-server, which runs through npx. Add an entry like this to ~/.claude.json, replacing the placeholder with your own key.
Get an API key at exa.ai. The skill needs Node.js and npm installed so that npx can run the server, and it needs internet access.
- Requires an Exa account and API key from exa.ai
- Requires Node.js/npm to run the MCP server
- Key is stored in the env block of ~/.claude.json
"exa-web-search": {
"command": "npx",
"args": ["-y", "exa-mcp-server"],
"env": { "EXA_API_KEY": "YOUR_EXA_API_KEY_HERE" }
}Trigger the skill with natural requests
You do not need to name the tools. The skill activates when you ask for current web information, code examples, company or competitor research, professional profiles, or deep research. Phrases such as "search for", "look up", "find", and "what's the latest on" are listed as triggers in the skill.
Be specific about the goal. A request that names a topic, a time range, and a source type gives the agent enough to choose the right tool.
- "Look up the latest news on EU AI regulation."
- "Find code examples for Rust error handling with the Result type."
- "Research Vercel's funding and valuation in 2026."
Use the tools: examples and parameters
Each tool takes a query and a few optional settings. web_search_exa returns 8 results by default; you can lower this with numResults. get_code_context_exa defaults to 5000 content tokens, which you can set between 1000 and 50000.
The examples below show the calls the skill documents. You can ask the agent for any of them in plain language.
- web_search_advanced_exa accepts includeDomains, excludeDomains, startPublishedDate and endPublishedDate.
- company_research_exa defaults to 5 results.
- crawling_exa reads a single URL and accepts tokensNum (default 5000).
web_search_exa(query: "latest AI developments 2026", numResults: 5)
get_code_context_exa(query: "Python asyncio patterns", tokensNum: 3000)
company_research_exa(companyName: "Anthropic", numResults: 5)
crawling_exa(url: "https://example.com/article", tokensNum: 5000)Tips for better results
Use the broad search for general questions and the filtered search when you need results from particular sites or dates. For focused code snippets, lower tokensNum to 1000 to 2000; for fuller context, use 5000 or more.
For company due diligence, combine company research with a filtered web search. Use crawling_exa to read a specific page that appeared in results. Deep research suits comprehensive topics that benefit from AI synthesis, and it runs in the background so you can keep working.
- Use web_search_exa for broad queries and web_search_advanced_exa for filtered ones
- Combine company_research_exa with web_search_advanced_exa for company analysis
- Use crawling_exa on URLs found in search results
- Related skills in the source: deep-research (firecrawl and exa together) and market-research
What people use it for
- Pulling current API documentation and code examples while writing code
- Researching a competitor's funding, news and market position before a meeting
- Finding recent news or web facts the agent may not know about
- Reading the full text of an article or page from a URL found in search results
- Starting a longer background research job on a technical topic and checking back later
Frequently asked questions
Is the exa-search skill free?
The skill is published under the MIT licence. Exa is an external service that requires an API key from exa.ai, and every search is an external call, so check Exa's pricing before you rely on it heavily.
What do I need to use the exa-search skill?
You need an Exa API key from exa.ai, Node.js and npm so the exa-mcp-server package can run through npx, and internet access. The Exa MCP server must be added to ~/.claude.json with your key.
Where do I install the exa-search skill?
Unzip the downloaded package into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
Why isn't the exa-search skill working?
The most likely cause is a missing or incorrect EXA_API_KEY, or the MCP server entry missing from ~/.claude.json. Confirm Node.js and npm are installed, since the server is started with npx -y exa-mcp-server.
Does the skill run code on my machine?
Yes. The skill runs scripts and uses the network, and it starts the exa-mcp-server package through npx. Review the files in the package before installing, and keep your API key out of shared projects.
How is this different from the agent's normal web search or another search skill?
The skill gives the agent Exa's neural search tools, including code context, company research, people search and deep research, instead of answering only from built-in knowledge. Its own documentation points to deep-research and market-research as related skills for broader workflows.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the exa-search skill if you are a developer who needs current code examples and API docs, or an analyst doing company and competitor research, and you are willing to set up an Exa account and API key. Skip it for tasks that need no live information, for looking up private individuals, or if you cannot use a paid external service.
Download the skill