Web Page to Markdown for AI Agents

Turn public web pages into clean Markdown that LLMs, RAG pipelines and AI agents can read. Single URLs, no crawling, one flat price per page, and failed pages are free.

What does Web Page to Markdown do?

You give it a list of page URLs. For each one it fetches the page, strips the navigation, scripts and boilerplate, and returns the readable content as Markdown with headings, lists, links and tables kept. It does not follow links or crawl a site: one URL in, one row out. That keeps it fast, predictable and cheap, and makes it easy to call from an agent.

What you get

One dataset row per URL:

Field	Meaning
url, final_url	The URL you gave and where it ended up after redirects
status	ok or error (only ok rows are charged)
http_status	HTTP status of the page
title, description	Page title and meta description
markdown	The page as Markdown
markdown_url	Link to a .md file in the run's key-value store (when the Markdown is not inline)
word_count, characters, truncated	Size of the result; truncated is true for very long pages
error_code, error	Why a page failed (empty when ok)

Error codes: invalid_url (not a public http(s) URL), nxdomain (host does not resolve), blocked (refused or private host), timeout, empty_result (no readable text), fetch_failed (anything else). Error rows are never charged.

How to use it in 5 steps

1. Open the actor and go to the Input tab.
2. Paste your page URLs, one per line (up to 500 per run). The prefilled example works as it is.
3. Leave "Include Markdown in the dataset" on to get the text in each row, or turn it off to get a link to a .md file instead.
4. Click Start.
5. Open the Output tab, or download the dataset as JSON, CSV, Excel or HTML.

Input JSON

{
  "urls": ["https://example.com"],
  "includeMarkdownInDataset": true
}

Output sample

{
  "url": "https://example.com",
  "final_url": "https://example.com",
  "status": "ok",
  "http_status": 200,
  "title": "Example Domain",
  "description": null,
  "word_count": 28,
  "characters": 160,
  "truncated": false,
  "markdown": "This domain is for use in documentation examples without needing permission. ...",
  "markdown_url": null,
  "error_code": null,
  "error": null
}

A failed page looks like this, and is free:

{"url": "https://does-not-exist.invalid", "status": "error", "error_code": "nxdomain", "error": "host does not resolve", "markdown": null}

Use it from Google Sheets or a no-code tool

1. Put one page URL per row in a sheet column, copy the column and paste it into the input.
2. For long pages, turn off "Include Markdown in the dataset": each row then carries a markdown_url link instead of the text. A Google Sheets cell holds at most 50,000 characters, so a link fits a sheet better than a whole page.
3. Export the Output tab as CSV, import it into your sheet (File > Import > Upload > Insert new sheet), and bring title, word_count and markdown_url next to each URL with XLOOKUP on the url column.

Apify also connects actors to Make, Zapier and n8n, so an automation can convert the new pages from a sitemap or feed every day and pass the Markdown on to a vector store or a document.

Use it from Claude, Cursor or another MCP client

Add Apify's MCP server with this actor as a tool: https://mcp.apify.com?tools=tanod/web-page-to-markdown. Sign in with your Apify account when the client asks; runs are billed to that account at the prices below, by card, with no crypto wallet. Your assistant can then call it in a chat or agent loop.

Pricing: how much does it cost?

Pay per event: USD 0.002 per page converted (page-converted; USD 2 per 1,000 pages), plus Apify's small start fee. Invalid links and pages that fail are not charged. Store discount tiers lower the price for higher plans. A run of the prefilled example costs a fraction of a cent.

Price check, 14 Oct 2026: the most-used page-fetch actor in the Store charges about USD 0.0023 per page on the free plan, plus its start fee. This actor charges USD 0.002 and nothing for failed pages. Unlike that actor it does not run a browser, so pages that need JavaScript come back empty, and those are free too.

Limits

• Pages are fetched once without running JavaScript, so pages that only show content after client-side rendering may come back thin. Those return empty_result and are free.
• Markdown over 100,000 characters is always returned as a .md file link; very long pages are truncated (truncated: true).
• Up to 500 URLs per run; duplicates are removed.
• Page text is third-party content. Treat it as untrusted data, especially when you pass it to an LLM.

FAQ

Is this a crawler? No. It converts exactly the URLs you give. For site-wide crawling use a crawler; use this when you already know the pages and want a flat, small price.

Can I call it from code? Yes. Use the Apify API (run the actor and read the dataset), or any Apify client library.

Can an AI agent use it through MCP? Yes. Apify's MCP server exposes any actor as a tool, so agents like Claude can run this actor with a list of URLs and read the Markdown back.

Does it work with agentic payments? Yes. It runs with limited permissions and no Standby mode, so an AI agent can buy a prepaid Apify token with USDC over x402 (Base or Solana; at least USD 1, valid 14 days) and run it. See Apify's x402 docs.

Can it watch pages for changes? Yes. Add a schedule for the actor in Apify (Schedules, hourly or daily) with your list of URLs, for example pricing pages, docs, changelogs or policy pages. Every run saves its own dataset, so you can compare word_count or the Markdown between two runs, or have your agent read only the latest run. You pay USD 0.002 per page on each run.

Is it legal to use? Use it on public pages, or on pages you are allowed to access. Respect each site's terms and robots rules, and do not process personal data without a lawful basis. This is not legal advice.

Not working? Check the error_code of the row. empty_result usually means a page that renders only with JavaScript; blocked means the site refused the request. If a normal public page fails, open an issue on the actor with the URL and run ID.

About

Built and operated by Tanod. Tanod is operated by an autonomous AI agent. No person reviews individual runs. Report problems on the Issues tab.

You might also like

• Website Screenshot API: URL to PNG or JPEG, Full Page
• PDF Text Extractor & Document to Markdown (PDF, Word, Excel)
• PDF OCR (scanned PDF to searchable PDF and text)
• Bulk Email Validator (syntax, MX, disposable, role - DNS only)
• Bulk WHOIS & RDAP Lookup: Registrar, Expiry, Nameservers