← Back to blog

3 Free Web Utility APIs for AI Agents: Screenshots, HTML to PDF & Readability

Your AI agent can browse the web — sort of. But what if it needs to see a page as a human would? Or grab a clean article without the navigation bar clutter? Or turn a web page into a PDF for archiving?

Agent Media Tools just shipped three new web utility APIs that fill exactly these gaps. They're free, they work with a single curl or Python call, and they're designed to slot right into an AI agent's tool belt.

The Three Endpoints

• 📸 Screenshot API — full-page PNG screenshot of any URL
• 📋 Article Readability — extract clean title, author, and body text from any article URL
• 📄 HTML to PDF — render any web page to a printable PDF document

All three share the same base URL: https://agentmediatools.com/api. You'll need a free API key from the dashboard.

1. 📸 Screenshot API — Full-Page Capture

Pass a URL, get back a full-width, full-height PNG screenshot. No viewport clipping, no half-captured pages — it scrolls the entire document and stitches the result.

curl

curl -X POST https://agentmediatools.com/api/screenshot \
  -H "x-api-key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com"}' \
  --output page.png

Python

import requests

resp = requests.post(
    "https://agentmediatools.com/api/screenshot",
    headers={"x-api-key": "YOUR_KEY"},
    json={"url": "https://example.com"}
)
with open("page.png", "wb") as f:
    f.write(resp.content)

Optional parameters: width (default 1280), height (default 0 = full page), delay (ms to wait for JS rendering).

2. 📋 Article Readability — Clean Text Extraction

Give it a news article or blog post URL, and it returns structured JSON with the title, author, publication date, featured image, and clean body text — stripped of ads, nav bars, and sidebars.

This is the same algorithm Mozilla's Readability uses behind Firefox's Reader View.

curl

curl -X POST https://agentmediatools.com/api/article \
  -H "x-api-key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://en.wikipedia.org/wiki/Artificial_intelligence"}'

Python

import requests

resp = requests.post(
    "https://agentmediatools.com/api/article",
    headers={"x-api-key": "YOUR_KEY"},
    json={"url": "https://en.wikipedia.org/wiki/Artificial_intelligence"}
)
article = resp.json()
print(article["title"])      # "Artificial intelligence"
print(article["text"][:200]) # Clean article text, no clutter

Returns: title, author, date, image (og:image), text (cleaned body), excerpt.

3. 📄 HTML to PDF — Page to Document

Render any URL as a proper PDF document. Perfect for archiving, shareable reports, or feeding printed-looking content into document processing pipelines.

curl

curl -X POST https://agentmediatools.com/api/html-to-pdf \
  -H "x-api-key: YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com"}' \
  --output page.pdf

Python

import requests

resp = requests.post(
    "https://agentmediatools.com/api/html-to-pdf",
    headers={"x-api-key": "YOUR_KEY"},
    json={"url": "https://example.com"}
)
with open("page.pdf", "wb") as f:
    f.write(resp.content)

Optional: format ("A4" or "Letter", default "A4"), landscape (bool).

Why This Matters for AI Agents

AI agents are great at pulling text from APIs, but the web wasn't built for APIs — it was built for humans. These three endpoints bridge that gap:

• Screenshot lets your agent see a page visually. Useful for verifying UI changes, monitoring dashboards, or capturing proof of content.
• Readability turns messy HTML into clean, LLM-friendly text. Feed the output directly into a summarizer, translator, or knowledge base.
• HTML to PDF turns live pages into portable documents. Great for "save this for later" workflows, report generation, and compliance archiving.

AI Agent Integration (MCP)

If you use Claude with MCP, all three tools are already exposed through the MCP server. Your Claude instance can take screenshots, extract articles, and generate PDFs on demand — no code required.

Rate Limits & Pricing

• Anonymous: 10 calls/day
• Free login: 25 calls/day
• Premium ($5/mo): 1,000 calls/day

All three endpoints count toward the same pool, so you can mix and match.