Markdown Link Extractor

List every link in a Markdown file

A long README, a knowledge-base article or a set of course notes can hold dozens of links: references to documentation, download pages, internal sections, email addresses and bare URLs pasted straight into the text. The Markdown Link Extractor finds them and lists each one with its anchor text, its address and its optional title. You get a readable table for a quick look and a JSON array you can copy or download for audits, migrations and link checks.

Two kinds of links are collected

The extractor runs two separate passes over the text outside code fences. Images are blanked out first, so they never appear in the results:

1. Markdown links in the form [text](url) or [text](url "title"). These come out with their text, URL and title. Relative paths, anchors such as #install, and mailto: addresses are all captured this way.
2. Bare web addresses starting with http:// or https:// anywhere else in the text, including autolinks like <https://example.com>. Markdown links found in the first pass are removed before this pass runs, so their URLs are not counted again. These come out with the URL only, and text set to null.

Each result is an object such as:

{"text": "Docs", "url": "https://example.com/docs", "title": "Read the docs"}

Within each stretch of text between code fences, the Markdown links are listed first and the bare addresses follow.

Each link is listed once

A link written as [Docs](https://example.com/docs) produces one entry, with the text "Docs", because the bare-address pass skips URLs that already belong to a Markdown link. A URL appears more than once only when your document really contains it more than once, for example two links to the same page with different anchor text. The tool does not merge those, so de-duplicate the JSON by URL if you want one row per address.

To find addresses that were pasted as bare URLs, look for entries with text set to null.

Other behaviour worth knowing

• Images are excluded. An image such as ![Logo](logo.png) is removed before either pass, so neither the image file nor its alt text is listed. Use the Markdown Image Extractor for images.
• Reference-style links such as [guide][1] are not captured as links. If the definition line [1]: https://example.com holds a web address, that address is picked up by the bare-URL pass, without text.
• Fenced code blocks are skipped, but inline code in single backticks is not, so a link shown as an example inside backticks is still listed.
• Titles must be in double quotes to be recognised.
• Angle brackets around a URL are trimmed from the result.

A sample run

Consider this paragraph: Read the [guide](https://example.com/guide), email [support](mailto:[email protected]), or visit https://example.com/status for uptime. ![Logo](https://example.com/logo.png) The extractor returns three entries. First come the two Markdown links: "guide" pointing to https://example.com/guide and "support" pointing to the mail address. Then comes the bare address https://example.com/status with no text. The guide URL is not repeated as a bare address, and the logo image is not listed at all.

Step-by-step guide

1. Paste your Markdown into Input Markdown, or press Paste in the header.
2. Check the "possible links" counter in the footer. It is a quick browser-side estimate of Markdown links only.
3. Press Extract. The button reads "Working" while the request runs.
4. Read the Extracted Links table, which shows Text, URL and Title for each entry. Web addresses, mail and phone links, root-relative paths and anchors are clickable and open in a new tab.
5. Copy the array with the copy icon (Copy JSON), or save it as markdown-links.json with the download icon (Download JSON).
6. Clear the input and results with the trash icon.

Server processing

The extraction is done on this site's server. When you press Extract your Markdown is sent there, both passes run, and the list returns to your browser. The tool does not visit or test any of the links. Each request accepts up to 200,000 characters; anything longer is rejected with an error message, so split very large documents first.

Where a link list helps

Link audits before publishing

Export the list, then open or check each URL. It is much easier to work through a column of addresses than to hunt for links in rendered text.

Domain and site migrations

When a product moves from one domain to another, the URL list shows every reference to the old domain that needs rewriting, including bare URLs that are easy to overlook.

Reviewing outbound references

Editors and SEO teams can see at a glance which external sites a page points to and whether the anchor text describes them well. Vague anchors such as "here" or "click this" stand out in the Text column.

Building a reading list or bibliography

Researchers and students can pull every source from a set of notes into one file, then format it however they need.

Checking internal navigation

Relative links and anchors show how a document set links to itself. Missing or misspelled paths are easier to spot in a sorted list.

Tips

• De-duplicate by URL after exporting if the same address is linked several times and you want one row per address.
• Write links in Markdown form rather than as bare URLs when you want anchor text in the export; bare addresses come out with text set to null.
• Avoid spaces in link targets. A URL containing a space is not matched by the Markdown link pattern; encode it as %20.
• Expect a bare URL followed by a period at the end of a sentence to include that period. Check URLs at sentence ends.

What not to expect

• No broken-link checking. The extractor lists links; it does not fetch them.
• No HTML link detection. Anchor tags written as <a href="..."> contribute only their bare http or https address, without text.
• No automatic de-duplication or sorting.
• No image listing; images are skipped.

Limitations

• Repeated links to the same address are each listed; there is no merging.
• Reference-style links lose their text.
• Bare addresses are matched only when they start with http or https; addresses like www.example.com on their own are missed.
• Input is capped at 200,000 characters and processed on the server.
• JSON is the only export format.

Related tools

To actually test whether each address responds, use the Markdown Broken Link Checker. If you need the links removed but their text kept, Markdown Strip Links does that, and the Markdown Image Extractor lists images on their own.

Frequently Asked Questions