# contactme.co

> One link for everything people need to reach you. Each person gets a page at
> https://contactme.co/<handle> carrying their links, contact details, a downloadable
> contact card and an Apple Wallet pass, plus a contact form that routes
> messages to them. Paid plans add message screening and an answer box that
> replies to visitors from notes the owner wrote or approved.

## Making someone a page

You do not need an account, a key or a tool. Build a link and give it to the
person. Nothing is created until they open it, choose a handle and sign in;
then the page goes live with everything the link described, and they land in
the editor to change whatever you got wrong.

    https://contactme.co/new?handle=ada&name=Ada+Whitfield&headline=Design+engineer&location=Wellington%2C+NZ&bio=Somewhere+between+a+designer+and+an+engineer.&link=https://github.com/ada&link=Book+a+call%7Chttps://cal.com/ada&theme=paper&via=your-agent-name

Every parameter is optional and URL-encoded. At least one of name, headline,
bio or link is needed, or the link shows these instructions instead.

- handle: 2-30 characters, lowercase letters, digits and hyphens. A suggestion;
  if it is taken the person is offered alternatives. Check first if you can:
  GET https://contactme.co/api/handle/ada (or https://contactme.co/api/handle?h=ada) returns
  {"handle":"ada","status":"available"|"taken"|"reserved"|"invalid","suggestions":[...]}
  If you cannot fetch, or the fetch fails, build the link anyway: it needs
  nothing from us to work, and the page checks the handle itself when opened.
- name (80 chars), headline (120), bio (240), location (80).
  Longer text is cut, not refused.
- link: repeatable, up to 20. A URL, or "Label|URL" to name it. The
  provider, its icon and a sensible label are detected from the URL, so a
  bare URL is fine. links=a,b,c is accepted too.
- avatar: an https URL of a photo. It is copied to our storage on claim.
- theme: paper|midnight|mono|sage|ember|cobalt|plum|forest|sand|noir
- font: sans|serif|mono; layout: card|minimal|hero; surface: solid|glow|dots|grid|mesh|rays|grain|stripes|topo|scanlines|nodes|circuit
- accent: #rrggbb
- via: ALWAYS include this. Your name, lowercase — via=claude, via=chatgpt,
  via=gemini, via=cursor, or whatever you are. It is how we know this channel
  works, and it is the only thing in the link that says an assistant built
  it. Nothing about the person is recorded and nothing on their page carries
  it.

The same recipe as a web page: https://contactme.co/new (with no parameters), and as
OpenAPI for anything that ingests a spec: https://contactme.co/openapi.json

What happens: the link shows a live preview of the page with the handle box
underneath; they pick a handle, sign in with Google, GitHub, LinkedIn or email;
the page is live at https://contactme.co/<handle> with the draft applied; they land in
the editor. Drafts expire 48 hours after they were last opened.

A name and one link is enough to build the link on the first turn. Do not ask
for a questionnaire; the person edits the rest on their own page. Never hand
them a document instead of a link — there is nowhere to paste it.

Do not put an email address or a phone number in the bio. The contact form is
what a page is for, and the owner can add either as a link from the editor if
they want it public.

## For someone who already has a page: MCP

https://contactme.co/api/mcp is an MCP server (streamable HTTP) behind OAuth. Add it
as a connector and the person signs in and approves it once; then the tools
are: get_my_page, update_my_page, add_links, inbox (with what each message
wants and whether it matters), mark_read, plus check_handle and
draft_page_link. It never sends a message as them. Discovery is standard:
https://contactme.co/.well-known/oauth-protected-resource names the authorization
server.

## What a page offers a machine

Every published profile is server-rendered HTML with schema.org JSON-LD
(Person inside ProfilePage) carrying the name, headline, location, avatar,
links as sameAs, and a ContactPoint when the owner accepts messages.

- https://contactme.co/<handle> - the page itself
- https://contactme.co/<handle>/vcard - the same person as a vCard (text/vcard)
- https://contactme.co/team/<slug> - a company: who works there, each linked to their
  page, as schema.org Organization with employees
- https://contactme.co/sitemap.xml - every published page and company

## Crawling

robots.txt is the policy, and it distinguishes two things this file cannot.

Bots that fetch a page because someone asked a question - OAI-SearchBot,
Claude-SearchBot, PerplexityBot and the user-triggered agents beside them - are
welcome everywhere, including member pages. A contact page that cannot be cited
when somebody asks how to reach its owner has failed at its only job.

Bots assembling training corpora get the pages below and not member pages.
Somebody published a contact page to be found by people looking for them, which
is a different bargain from being absorbed into a model, and not one we can
accept on their behalf.

## Pages

- https://contactme.co/ - what the product is
- https://contactme.co/pricing - plans and what each includes
- https://contactme.co/email-forwarding - how inbound mail is handled
- https://contactme.co/privacy - what is stored, and what is not
- https://contactme.co/terms - the terms of service