Haunt API: Laser-focused Web Extraction for AI Agents

[Image: Haunt API] FeaturesPricingAgentsDocsServicesBlogGet API key

agent extraction array

Web extraction with laser accuracy.

Haunt gives AI agents a clean way to read the web. Use MCP or REST, send a URL and a plain-English prompt, then get structured JSON with trace evidence.

Get API keyRead the docs

100 requests free MCP in 30 seconds Trace-backed JSON

Haunt extraction consoleonline

Target URL news.ycombinator.comqueued

Prompt top stories, titles, pointsparsed

Output structured JSONlocked

Trace fetch, confidence, quotaincluded

mcp quickstart

One install. Every AI tool.

Claude, Cursor, Windsurf, Cline, or direct HTTP. Same core extraction engine.

claude_desktop_config.json MCPPythoncURLJS

// Add to Claude Desktop, Cursor, or any MCP client
{
 "mcpServers": {
 "haunt": {
 "command": "npx",
 "args": ["@hauntapi/mcp-server"],
 "env": { "HAUNT_API_KEY": "PASTE" }
 }
 }
}

import requests

resp = requests.post(
 "https://hauntapi.com/v1/extract",
 headers={"x-api-key": "PASTE"},
 json={"url": "https://news.ycombinator.com", "prompt": "Get top stories with titles and points"}
)
print(resp.json())

curl -X POST https://hauntapi.com/v1/extract \
 -H "Content-Type: application/json" \
 -H "x-api-key: PASTE" \
 -d '{"url":"https://news.ycombinator.com","prompt":"Get top stories"}'

const resp = await fetch("https://hauntapi.com/v1/extract", {
 method: "POST",
 headers: { "Content-Type": "application/json", "x-api-key": "PASTE" },
 body: JSON.stringify({ url: "https://news.ycombinator.com", prompt: "Get top stories" })
});
console.log(await resp.json());

why haunt

Built for agents that need the web, not another parser to babysit.

Selectors rot. Browser scripts get blocked. Raw HTML is not the product. Haunt turns pages into usable data.

prompt

Natural language extraction

Describe the data you want in plain English. Haunt returns structured fields instead of making you maintain brittle selectors.

fallback

CAPTCHA-aware extraction

When pages need human verification, Haunt stops cleanly and sends structured verification flags instead of pretending the challenge is solved.

rest

One POST request

No browser farm. No Puppeteer wiring. Send a URL and prompt, get clean JSON back.

auth

Authenticated scraping

On Pro and Scale, send cookies or headers when the data lives behind your own login wall.

batch

Batch processing

Extract from multiple URLs when your agent needs a broader scan without another loop of glue code.

json

Structured output

Arrays, objects, numbers, strings, and trace metadata. Ready for your app, database, or agent pipeline.

trace evidence

Real response. Evidence trace included.

Clean JSON plus a safe trace showing how Haunt fetched and trusted the result.

Success response, 743msconfidence 88%

// POST /v1/extract
{
 "success": true,
 "data": { "main_heading": "Example Domain" },
 "credits_remaining": 99,
 "trace": {
 "fetch": { "source": "direct", "status_code": 200 },
 "extraction": { "mode": "full_structured", "confidence": 0.88 },
 "summary": "Fetched with direct; returned full_structured at 88% confidence."
 }
}

CAPTCHA wall responsehuman verification required

// POST /v1/extract
{
 "success": false,
 "error": "CAPTCHA required: the target page requires human verification.",
 "error_code": "captcha_required",
 "captcha_provider": "hcaptcha",
 "requires_human_verification": true,
 "url": "https://example.com"
}

pricing matrix

Simple pricing.

Start free. Paid plans now give you room to build. No hidden fees.

£ GBP$ USD€ EUR

Free

£0 100 requests / month

• AI-powered extraction
• Natural language prompts
• JSON output
• Fallback fetching
• Authenticated scraping
• Priority support Get started

Starter

£19/mo 5,000 successful requests / month

• AI-powered extraction
• Natural language prompts
• JSON output
• Fallback fetching
• Authenticated scraping (Pro+)
• Priority support Start free trial

Pro

£49/mo 25,000 successful requests / month

• AI-powered extraction
• Natural language prompts
• JSON output
• Fallback fetching
• Authenticated scraping
• Priority support Start free trial

Scale

£99/mo 75,000 successful requests / month

• AI-powered extraction
• Natural language prompts
• JSON output
• Fallback fetching
• Authenticated scraping
• Priority support
• Batch processing
• Custom rate limits Start free trial

No automatic overage billing. Failed extractions do not count. Need more than Scale? Contact support before you hit the limit.

use cases

Built for agent workflows that need clean data.

MCP-powered agents

Give Claude, Cursor, or your own agent web extraction capabilities without writing a new scraper.

AI and LLM pipelines

Feed clean web data into research agents, RAG systems, support tools, and app workflows.

Research and data science

Collect structured web datasets without building custom parsers for every target site.

Price monitoring

Track product pages and competitors without maintaining selectors every week.

Lead enrichment

Extract company facts, contact routes, service descriptions, and proof points from websites.

News aggregation

Pull headlines, article summaries, and source metadata for monitoring and curation systems.

faq

Quick answers. No fog.

How is this different from BeautifulSoup?

BeautifulSoup needs selectors that break when sites change. Haunt uses AI extraction, so you describe the data you want and get structured JSON back.

Does it handle difficult or protected public pages?

Haunt tries the appropriate fetch path where supported. Interactive CAPTCHA, paywalls, and login flows are not guaranteed.

What if extraction fails?

Failed extractions do not count against quota. You get a structured error instead of silent junk.

Can I scrape behind a login?

Yes, on Pro and Scale. Free and Starter are public-page extraction only.

Is there a Python SDK?

Yes. Use pip install hauntapi, the MCP server, Postman, or any HTTP client.

Do you offer refunds?

Start free with 100 requests/month. If a paid plan is wrong for your use case, contact support within 14 days.

api key

Get your API key.

Free. No credit card. Start extracting in 30 seconds. Get free API key

Your API key: ``

Save this key. Check the docs to get started.

Haunt API, web extraction with teeth.Privacy · Terms · Docs · [email protected]