Tapmy.link for AI agents
Tapmy.link is a free link-in-bio service where every page lives on its own subdomain. A program can register, build and publish a page for a person in one HTTP call, then hand the person a link that makes the account theirs. This documentation is written for that program. Every page here is also available as Markdown: add `.md` to its address.
openapi.json/mcpskill.mdllms.txtllms-full.txtagent-card.json
Start here
What Tapmy.link is, what a page contains, what it costs, and what a program is allowed to do here.
Check a name, create the page, verify it. Copy-paste curl for each step, with the exact response shape.
What to do if you can only GET, or have no network at all: compose the page JSON and hand the person a one-click link to /new that creates it in their browser.
The two kinds of account a program can create, what a claim link does, and what happens to pages nobody claims.
Bearer keys: where they come from, how to send them, what they can and cannot do, how the owner revokes them.
Building a page
Every field on a page, its type and limits, and which endpoint sets it.
The six block types with the minimal JSON for each, every optional field, and what the templates do with them.
The 35 platform ids accepted in social and socialRow blocks, with the URL shape each expects. Generated from the code.
The ten templates, what each looks like, and which kind of page it suits. Switching never loses content.
How to put a photo on a page when all you have is a URL: the upload route, storage keys, kinds and size caps.
Give a link a path on the owner's subdomain, count taps on it, and send iPhone and Android to different stores.
How links to OnlyFans, Fansly and similar are handled, what a program must set, and what never appears in the page.
The twenty interface locales, how the account language and the page language differ, and which code to send where.
Copy the links from a Linktree, Taplink, Beacons or any public link page, or the photo, name and bio from Instagram.
Verify and hand over
Read a public page back as Markdown, check what a visitor will see, and confirm the page is actually published.
Views, unique visitors and taps for 7 and 30 days, with a daily series — what an agent can report back to its owner.
What to tell the person after the page exists: the two links, what the claim link does, and a message you can send verbatim.
What a program should and should not do with a page it did not pay for and does not own.
Reference
Every HTTP endpoint a program can call, with auth, purpose and a link to the page that explains it. The machine-readable form is /openapi.json.
The error envelope, every code the API can return, and the hint that accompanies each. Generated from the code.
Every limit a program can hit, how it is signalled, and the sweep that removes pages nobody claimed or visited.
The Model Context Protocol endpoint, its tools, how to connect from Claude, Cursor and other clients, and the one rule about keys.
A ready procedure for coding agents (Claude Code, Codex, Cursor): when to use Tapmy, the steps, the guardrails.
Dated changes to the agent-facing API, the MCP server and this documentation.
Recipes
A complete request body for a musician: the newest release first, one row of streaming marks, dates, merch, press.
A complete request body for a place: reserve first, then menu, directions, hours, delivery, and a chat button.
A complete request body for a photographer: portfolio first, booking, the current offer, prints.
A complete request body for a coach: one clear offer, a booking link, proof, a way to talk.
A complete request body for a shop: the current drop, categories as cards, delivery info, support chat.
A complete request body for a product or book launch: one page, one action, a countdown of steps, everything dated.