Exa Search

Exa Web Search

Use this skill to search the web via the Exa API using curl. No Node.js or Python needed — just shell.

Quick Start

curl -s -X POST 'https://api.exa.ai/search' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "YOUR QUERY HERE",
    "type": "auto",
    "num_results": 10,
    "contents": {
      "highlights": {
        "max_characters": 4000
      }
    }
  }'

Parameters

Common Search Patterns

General web search (default)

curl -s -X POST 'https://api.exa.ai/search' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "Next.js 15 app router data fetching",
    "type": "auto",
    "num_results": 5,
    "contents": {
      "highlights": { "max_characters": 4000 }
    }
  }'

Full page text (for code/docs you need in full)

curl -s -X POST 'https://api.exa.ai/search' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "Prisma schema relations one-to-many",
    "type": "auto",
    "num_results": 3,
    "contents": {
      "text": { "max_characters": 20000 }
    }
  }'

News / recent events

curl -s -X POST 'https://api.exa.ai/search' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "OpenAI GPT-5 release",
    "type": "auto",
    "category": "news",
    "num_results": 5,
    "contents": {
      "highlights": { "max_characters": 4000 }
    }
  }'

Deep research (thorough, slower)

curl -s -X POST 'https://api.exa.ai/search' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "query": "vector database comparison 2025 performance benchmarks",
    "type": "deep",
    "num_results": 10,
    "contents": {
      "highlights": { "max_characters": 4000 }
    }
  }'

Fetch content from known URLs

curl -s -X POST 'https://api.exa.ai/contents' \
  -H 'x-api-key: 69b2e0ee-f92a-48ab-9233-cc8a750a1e50' \
  -H 'Content-Type: application/json' \
  -d '{
    "urls": ["https://docs.example.com/api"],
    "text": { "max_characters": 20000 }
  }'

Reading Results

The response is JSON. Key fields per result:

.results[].title       — page title
.results[].url         — source URL
.results[].highlights  — relevant excerpts (array of strings)
.results[].text        — full text (only if you requested "text" contents)
.results[].publishedDate — ISO date string

Pretty-print with jq:

curl -s ... | jq '.results[] | {title, url, highlights}'

Extract just titles and URLs:

curl -s ... | jq -r '.results[] | "\(.title)\n\(.url)\n"'

Domain Filtering (optional)

Focus on authoritative sources when needed:

{
  "query": "...",
  "includeDomains": ["github.com", "docs.python.org"],
  "excludeDomains": ["pinterest.com", "quora.com"]
}

Environment Variable (recommended)

To avoid hardcoding the key in commands:

export EXA_API_KEY="69b2e0ee-f92a-48ab-9233-cc8a750a1e50"

curl -s -X POST 'https://api.exa.ai/search' \
  -H "x-api-key: $EXA_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{ "query": "...", "type": "auto", "num_results": 5, "contents": {"highlights": {"max_characters": 4000}} }'

Common Mistakes to Avoid

• ❌ "livecrawl": "preferred" → deprecated, use "maxAgeHours": 0 instead
• ❌ "useAutoprompt": true → deprecated, remove it
• ❌ "highlights" at top level of /search → must be inside "contents": { ... }
• ❌ "includeUrls" → doesn't exist, use "includeDomains"
• ❌ Both "text" and "highlights" in same request → pick one