Research / Constructs / Links

Markdown links in five publishing editors

One Markdown fixture, five real publishing editors, tested 2026-08-24. For links the observed result was untested in LinkedIn; a manual step in WordPress, Substack, Beehiiv, and X. Every row below is a rule with its own page and its own dated record.

Matches link_absolute, link_relative, link_anchor, link_local, link_mailto, link_unsafe, link_resolved. Fixture markdown-portability-test-v1. Tested 2026-08-24. How these tests are run.

What happens, per destination

Manual steps this creates

• LinkedIn: Check this part of the output yourself before posting. (The focused standard-post test did not isolate these constructs, so no reliable claim exists yet.)
• WordPress: Add an HTML anchor to each target heading, or remove the link. (The tested editor kept the href but generated no matching heading IDs.)
• WordPress: Add a base URL in Source context, or fix each link in the editor. (Without a confirmed site base URL there is no safe way to know where these point.)
• Substack: Add a base URL in Source context, or make each link absolute. (Substack rebound tested relative links onto its own publication path.)
• Beehiiv: Add a base URL in Source context, or fix each link in the editor. (Without a confirmed base URL there is no safe way to know where these point.)
• X: Add a base URL in Source context, or make each link absolute before copying. (A post is not a page of your site. Without a confirmed site base URL the raw path is pasted as ordinary text and points nowhere.)
• LinkedIn: Add a base URL in Source context, or make each link absolute before copying. (A feed post is not a page of your site. Without a confirmed site base URL the raw path is pasted as ordinary text and points nowhere.)
• WordPress: Replace it with an ordinary https link, or delete the text. (Only http, https, and mailto targets are carried. A javascript:, data:, or protocol-relative target is never copied into a destination editor.)
• Substack: Replace it with an ordinary https link, or delete the text. (Only http, https, and mailto targets are carried. A javascript:, data:, or protocol-relative target is never copied into a destination editor.)
• Beehiiv: Replace it with an ordinary https link, or delete the text. (Only http, https, and mailto targets are carried. A javascript:, data:, or protocol-relative target is never copied into a destination editor.)
• X: Replace it with an ordinary https link, or delete the text. (Only http, https, and mailto targets are carried. A javascript:, data:, or protocol-relative target is never copied into a destination editor.)
• LinkedIn: Replace it with an ordinary https link, or delete the text. (Only http, https, and mailto targets are carried. A javascript:, data:, or protocol-relative target is never copied into a destination editor.)
• WordPress: Check each rewritten target opens the page you meant. (Resolution is plain URL resolution against that base; no slug, extension, or redirect is guessed. Local media still needs an upload.)
• Substack: Check each rewritten target opens the page you meant. (Resolution is plain URL resolution against that base; no slug, extension, or redirect is guessed. Local media still needs an upload.)
• Beehiiv: Check each rewritten target opens the page you meant. (Resolution is plain URL resolution against that base; no slug, extension, or redirect is guessed. Local media still needs an upload.)
• X: Check each rewritten target opens the page you meant. (Resolution is plain URL resolution against that base; no slug, extension, or redirect is guessed. Local media still needs an upload.)
• LinkedIn: Check each rewritten target opens the page you meant. (Resolution is plain URL resolution against that base; no slug, extension, or redirect is guessed. Local media still needs an upload.)
• Substack: Check each in-page link after pasting. (Substack rewrote the tested anchor onto its draft URL.)
• Beehiiv: Check each in-page link in the preview. (The tested platform rewrote the anchor onto its own draft-preview URL.)

Limits of what was tested

• Each result describes the surface named beside it, on the date named beside it, and nothing wider.
• A result marked unknown means no dated test covers it. It is not a prediction.
• Rendering is not the same across platform versions, themes, plugins, or email clients.

Prepare a draft

• Check your own draft against all five destinations
• Markdown to WordPress, handoff mode: gutenberg paste
• Markdown to Substack, handoff mode: rich html clipboard
• Markdown to Beehiiv, handoff mode: clean markdown paste
• Markdown to X, handoff mode: plain text
• Markdown to LinkedIn, handoff mode: plain text

Related

• The 15 rules behind this page: BEE-ANCHOR-01, BEE-LINK-01, BEE-LINK-02, CORE-LINK-01, CORE-LINK-02, LI-LINK-01, LI-LINK-02, LI-SPECIAL-01, SUB-ANCHOR-01, SUB-LINK-01, SUB-LINK-02, WP-ANCHOR-01, WP-LINK-01, X-LINK-01, X-LINK-02
• Every construct, side by side
• The full five-destination matrix