Webpage to Markdown Converter for AI Agents

How It Works

Paste a public article, documentation page, or other webpage URL, or switch to HTML paste when you already have source markup. The parser locates the readable body, keeps headings, paragraphs, lists, tables, code, images, links, and source information, and removes navigation, ads, comments, recommendations, and repeated page furniture. Review the content boundary, source URL, and ending before copying or downloading Markdown. For a one-page task, this short path avoids sending the user through a separate landing page before the conversion starts. If the source is already visible in a browser, compare the displayed page with the generated body and use the failure message when an access boundary prevents a complete result.

Real Output, Not a Promise

A simple article should keep its title, sections, body, and source. A technical blog with a table and a command should keep the table header, row relationships, code fence, and language hint. Chinese content should remain in its source language even when the interface language changes. HTML to Markdown is about semantic structure, not a visual copy of every CSS rule.

What Gets Preserved

Headings and hierarchy, tables, code blocks, images with alt text and original URLs, links, and source metadata are useful after conversion. Navigation, ads, comments, recommendations, popups, cookie banners, and repeated page modules usually belong to the page shell and are removed. This gives a reader or AI agent a smaller and more traceable document. Before delivery, compare the first heading, one table or code block, a few links, and the final paragraph with the source. That review catches incomplete dynamic content without pretending that Markdown reproduces the original CSS, widgets, or private session.

Built for AI Agents and Developers

The API supports programmatic conversion and batch jobs. MCP lets compatible AI clients discover and call the same material tools. CLI fits terminals, scripts, and CI pipelines. The browser extension follows a page already rendered in the current browser. A reliable integration records the input URL, response status, parser message, source data, and whether a human review is needed.

How Herdown Compares

Firecrawl is strongly oriented toward developer web crawling, rendering, and job orchestration. Jina AI is strongly oriented toward remote reading, search, and content extraction. Herdown combines single-page URL conversion, local document tools, multi-page Website to Markdown, API, MCP, and CLI in a material-preparation workflow that emphasizes source retention, explicit failures, and a clear local-file boundary. Choose Firecrawl when a remote crawl job and endpoint controls are the priority. Choose Jina AI when a remote reader or search response is the priority. Choose Herdown when a person needs to review one public page, local documents, or a separate multi-page crawl with the same source and failure conventions.

Limitations

Login walls, paywalls, CAPTCHA screens, anti-bot rules, timeouts, and client-rendered pages can make a remote result incomplete. The service does not grant access to private pages or promise to bypass access controls. When a page is already visible in your browser, use the extension or paste HTML that you are allowed to access. Scanned material needs local OCR rather than a text parser.

Why Markdown for AI

HTML contains presentation tags, scripts, and page chrome that an agent does not need. Markdown makes the order and boundary of information easier to inspect. Headings, lists, tables, code fences, and links give retrieval and citation workflows recognizable structure, and reviewed Markdown can move into a RAG pipeline, prompt, repository, notes system, or publishing workflow. In a RAG pipeline, cleaner chunks reduce irrelevant context. In a prompt, visible headings and code fences make the requested material easier to follow. In documentation and publishing, a reviewed Markdown source can be reused without asking every downstream tool to interpret page CSS.

From Input to Delivery

Use URL to Markdown for one public page, Website to Markdown for a domain, starting URL, or sitemap with multiple pages, and Markdown Viewer to review the result. Check headings, tables, code, links, images, source metadata, and the ending before delivery.

Tested output examples

These examples come from real public-page conversions and a local HTML paste test. They show how titles, body text, links, code fences, and Chinese structure enter Markdown. Each example keeps both the input and output visible, so the result can be checked rather than accepted as an abstract promise.

Simple public page

https://example.com/

Example Domain

This domain is for use in documentation examples without needing permission. Avoid use in operations. [Learn more](https://iana.org/domains/example)

The title, body, link, and source can be checked directly, while the page shell is absent. The source page is intentionally short, so the output stays short instead of inventing paragraphs to make the result look larger.

Technical page code excerpt

https://developers.cloudflare.com/workers/get-started/guide/

### 1. Create a new Worker project

Open a terminal window and run C3 to create your Worker project.

```bash
npm create cloudflare@latest -- my-first-worker
```

The heading, explanation, commands, and code fence remain separate, so a developer can check the language hint and copy boundary. This is a structural excerpt, not a claim that the original page layout has been reproduced.

Chinese HTML paste fixture

# 转换测试

## 输入与输出

网页正文会保留标题、段落和来源链接。

This controlled HTML test verifies that Chinese headings, paragraphs, and levels remain in the source language. It is clearly labelled as a local fixture rather than presented as a remote Chinese page conversion.

Homepage FAQ

Is Herdown free to use?

Yes. The homepage offers a displayed free allowance for public URL conversion without sign-up. The allowance is shared with the API and is not an unlimited crawling promise.

Can I convert a webpage to Markdown online without installing anything?

Yes. Paste a public webpage URL in the browser and run the conversion online. No installation or API key is required for the normal free entry path.

Do you have a Webpage to Markdown Chrome extension?

Yes. It is useful when the current browser has already rendered the page or local extraction is a better boundary.

Can I convert HTML to Markdown?

Yes. Switch to HTML paste when you already have source markup.

Why can a page fail?

Login walls, paywalls, CAPTCHA screens, anti-bot rules, dynamic rendering, timeouts, or missing readable content can affect results.

Can I convert a webpage to Markdown with Python?

Yes. Python developers can call the REST API and save the returned Markdown locally.

Is there a webpage to Markdown example on GitHub?

The Herdown repository contains the open-source CLI, MCP, SDK, and developer documentation entry points.

What is the difference between Webpage to Markdown and Website to Markdown?

Webpage to Markdown converts one public URL into one Markdown result. Website to Markdown crawls multiple pages from a domain, starting URL, or sitemap.

Will my content be saved on your server?

Local HTML and document workflows run in the browser. Public URL conversion requests the submitted address to produce a result, but Herdown is not a content-hosting or knowledge-base service.