agentleFS
agenticlukeagenticluke/exa-research-plus/skill/SKILL.md

exa-search

Skill0 starsChanged 2 months agoMITView on GitHub

Search the web, code, companies, and people with Exa MCP. Use when the user asks for current news, facts, code samples, API docs, company details, expert profiles, or deep web research.

  • Reads credentials

66 other projects use a version of this, 18 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use deepxiv, figure-description and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • raja21068Skill

    deepxiv

    Search and progressively read open-access academic papers through DeepXiv. Use when the user wants layered paper access, section-level reading, trending papers, or DeepXiv-backed literature retrieval.

    Also in 4 projects that use this

  • raja21068Skill

    figure-description

    Process user-provided patent figures and generate formal drawing descriptions. Use when user says \"附图处理\", \"figure description\", \"附图说明\", \"drawings description\", or wants to describe patent figures with reference numerals.

    Also in 5 projects that use this

  • tpavanipradeepSkill

    bun-runtime

    Bun as runtime, package manager, bundler, and test runner. When to choose Bun vs Node, migration notes, and Vercel support.

    Also in 22 projects that use this

  • niloykumarbarmanSkill

    agent-sort

    Build an evidence-backed ECC install plan for a specific repo by sorting skills, commands, rules, hooks, and extras into DAILY vs LIBRARY buckets using parallel repo-aware review passes. Use when ECC should be trimmed to what a project actually needs instead of loading the full bundle.

    Also in 10 projects that use this

  • junimnjwSkill

    ralphinho-rfc-pipeline

    RFC-driven multi-agent DAG execution pattern with quality gates, merge queues, and work unit orchestration.

    Also in 9 projects that use this

  • mryeastcodeSkill

    everything-claude-code-conventions

    Development conventions and patterns for everything-claude-code. JavaScript project with conventional commits.

    Also in 8 projects that use this

  • +2 more from raja21068

The file

Show its rules and text

We split this skill into the rules it gives an agent

34 rules. 0 are common, 29 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 64 edited versions we read keep these.

  • 29

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use Exa MCP to find web pages, code, company facts, and people.

    line 11
  2. Distinctive:

    Use this skill when the user asks you to:

    line 15
  3. Distinctive:

    Do web research for a coding task

    line 22
  4. Distinctive:

    Do not use it when the answer is already in the files the user gave you.

    line 26
  5. Distinctive:

    Do not guess a tool name or field.

    line 37
  6. Very specific:

    Add the Exa MCP server to ~/.claude.json:

    line 41
  7. Distinctive:

    Never print, save, or share the API key in an answer.

    line 55
  8. Distinctive:

    If the server or key is missing, say what is missing.

    line 57
  9. Distinctive:

    Use this tool for news, facts, companies, people, and broad web search.

    line 63
  10. Distinctive:

    Use this tool for code samples and docs from sites such as GitHub, Stack Overflow, and API doc sites.

    line 84
  11. Distinctive:

    Use 1000 to 2000 for a small code sample.

    line 100
  12. Distinctive:

    Start with three to five results.

    line 106
  13. Distinctive:

    Check dates, source names, and page text.

    line 107
  14. Distinctive:

    Search again with better terms if the results are weak.

    line 108
  15. Distinctive:

    Do not treat a name match as proof that two profiles are the same person.

    line 156
  16. Distinctive:

    Search for fresh facts:

    line 182
  17. Very specific:

    Check that the code fits Node.js 22.

    line 200
  18. Distinctive:

    Add one small code sample.

    line 202
  19. Distinctive:

    If there are no results, make the query shorter and remove filters.

    line 207
  20. Distinctive:

    If results are off topic, add a key term, site name, date, or quoted phrase.

    line 208
  21. Distinctive:

    If sources disagree, show the difference.

    line 209
  22. Distinctive:

    If a page is old, do not use it as proof of a current fact.

    line 210
  23. Distinctive:

    If live crawl fails, use the search text and state that the full page was not checked.

    line 211
  24. Distinctive:

    If a tool rejects a field, remove that field and use only fields the server lists.

    line 212
  25. Distinctive:

    If the account hits a limit, stop and tell the user.

    line 213
  26. Distinctive:

    If code comes from an unknown source, treat it as an example.

    line 214
  27. Distinctive:

    Do not search for private data, keys, passwords, or login details.

    line 215
  28. Distinctive:

    Do not claim that search results are complete.

    line 216
  29. Very specific:

    Use web_search_exa for broad web facts and fresh news.

    line 220
  30. Very specific:

    Use get_code_context_exa for code and API use.

    line 221
  31. Distinctive:

    Add a site name when the user wants an official source.

    line 222
  32. Distinctive:

    Use quoted words only when the exact phrase matters.

    line 223
  33. Very specific:

    Use site: or intitle: only if Exa supports them well for the query.

    line 224
  34. Distinctive:

    Keep the first search small.

    line 225

“Common” is measured: kept by at least 6 in 10 of the 64 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Exa Search

Credit: This skill comes from ECC. Keep this credit when you copy or change the skill.

Use Exa MCP to find web pages, code, company facts, and people.

When to use it

Use this skill when the user asks you to:

  • Search the web
  • Find recent news or facts
  • Find code samples or API docs
  • Study a company or its rivals
  • Find experts in a field
  • Do web research for a coding task
  • "Search for," "look up," or "find" something
  • Explain the latest news about a topic

Do not use it when the answer is already in the files the user gave you.

Before you search

Exa can change its tool names, fields, and account limits. Check the tools that the Exa MCP server lists before you call them. Use the exact tool names and fields shown there.

This skill knows these tools:

  • web_search_exa
  • get_code_context_exa

Do not guess a tool name or field. If a field is not listed by the server, leave it out.

Setup

Add the Exa MCP server to ~/.claude.json:

"exa-web-search": {
  "command": "npx",
  "args": ["-y", "exa-mcp-server"],
  "env": {
    "EXA_API_KEY": "YOUR_EXA_API_KEY_HERE"
  }
}

Get an API key from exa.ai.

Never print, save, or share the API key in an answer.

If the server or key is missing, say what is missing. Do not make up search results.

Tools

web_search_exa

Use this tool for news, facts, companies, people, and broad web search.

web_search_exa(
  query: "latest AI news",
  numResults: 5
)

Common fields:

FieldTypeDefaultUse
querystringrequiredWhat to search for
numResultsnumber8How many results to get
typestringautoSearch mode
livecrawlstringfallbackAsk for fresh page text when needed
categorystringnoneNarrow the search, such as company or research paper

get_code_context_exa

Use this tool for code samples and docs from sites such as GitHub, Stack Overflow, and API doc sites.

get_code_context_exa(
  query: "Python asyncio task group example",
  tokensNum: 3000
)

Common fields:

FieldTypeDefaultUse
querystringrequiredThe code or API topic
tokensNumnumber5000How much text to get, from 1000 to 50000

Use 1000 to 2000 for a small code sample. Use 5000 or more for a broad topic.

Search steps

  1. Turn the request into a clear search query.
  2. Pick the right tool.
  3. Start with three to five results.
  4. Check dates, source names, and page text.
  5. Search again with better terms if the results are weak.
  6. Answer the user in your own words.
  7. Link to the pages that support the answer.
  8. Mark any guess or weak claim as unsure.

For current news, favor pages with a clear date. Check when the event took place, not only when the page was posted.

For a key fact, try to find two good sources. Favor first-hand sources, such as official docs, company pages, public records, and research papers.

Search patterns

Quick web search

web_search_exa(
  query: "Node.js 22 new features",
  numResults: 3
)

Code search

get_code_context_exa(
  query: "Rust Result error handling examples",
  tokensNum: 3000
)

Company search

web_search_exa(
  query: "Vercel funding and value 2026",
  numResults: 3,
  category: "company"
)

Person search

web_search_exa(
  query: "AI safety researchers at Anthropic",
  numResults: 5
)

Do not treat a name match as proof that two profiles are the same person. Check the job, group, location, and recent work.

Deep tech search

web_search_exa(
  query: "WebAssembly component model status and use",
  numResults: 5
)

get_code_context_exa(
  query: "WebAssembly component model code examples",
  tokensNum: 4000
)

Concrete example

User request:

Find the main changes in Node.js 22 and show one code sample.

Steps:

  1. Search for fresh facts:
web_search_exa(
  query: "Node.js 22 main changes official release notes",
  numResults: 4
)
  1. Search for code:
get_code_context_exa(
  query: "Node.js 22 new feature code example official docs",
  tokensNum: 2000
)
  1. Check that the code fits Node.js 22.
  2. Give a short list of the main changes.
  3. Add one small code sample.
  4. Link to the release notes and docs.

Edge cases

  • If there are no results, make the query shorter and remove filters.
  • If results are off topic, add a key term, site name, date, or quoted phrase.
  • If sources disagree, show the difference. Do not hide it.
  • If a page is old, do not use it as proof of a current fact.
  • If live crawl fails, use the search text and state that the full page was not checked.
  • If a tool rejects a field, remove that field and use only fields the server lists.
  • If the account hits a limit, stop and tell the user. Do not repeat calls in a loop.
  • If code comes from an unknown source, treat it as an example. Check it before you tell the user to run it.
  • Do not search for private data, keys, passwords, or login details.
  • Do not claim that search results are complete. Some pages may be missing.

Tips

  • Use web_search_exa for broad web facts and fresh news.
  • Use get_code_context_exa for code and API use.
  • Add a site name when the user wants an official source.
  • Use quoted words only when the exact phrase matters.
  • Use site: or intitle: only if Exa supports them well for the query.
  • Keep the first search small. Search again only when needed.

Related skills

  • deep-research: Uses Firecrawl and Exa for a full research flow.
  • market-research: Studies a market to help with a business choice.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.