# Exa Search API

Neural search engine & structured content extractor for AI agents

Exa is an embedding-powered web search engine designed for LLMs, delivering clean semantic snippets and structured page extraction.

## Overview

- **Website**: https://exa.ai
- **Docs**: https://docs.exa.ai
- **GitHub**: https://github.com/exa-labs/exa-py
- **License / Pricing**: Freemium
- **Directory Entry**: https://agentaccess.dk/enablers/exa

## About

Exa replaces traditional keyword search engines with semantic vector retrieval across the live web. AI agents can search using natural language or target URL embeddings, retrieving concise highlights and structured markdown formatted for context windows.

## Key Features

- Neural semantic search tuned for LLM context windows
- Domain filtering (e.g. restrict to .dk domains and Danish government sites)
- Sub-second highlight extraction with zero boilerplate
- Official Python and TypeScript SDKs + MCP server integration

## Danish System Application Recipes

### Neural Discovery of Danish Government Initiatives
Query Exa for "Danish green energy transition subsidies" restricted to `includeDomains: ["ens.dk", "kefm.dk"]` to retrieve clean Markdown highlights.

## Quickstart

```typescript
import Exa from 'exa-js'

const exa = new Exa(process.env.EXA_API_KEY)
const result = await exa.searchAndContents(
  'Danish energy infrastructure datasets',
  {
    type: 'neural',
    includeDomains: ['energinet.dk', 'ens.dk'],
    highlights: true,
  }
)

console.log(result.results)
```

---
Source: [AgentAccess Enablers](https://agentaccess.dk/enablers/exa)