OpenTree DashboardSign in

$ date -I # 2026-07-20

Making an agent's site public — and why private is the better default

The literal question has a one-word answer: yes. The reason it keeps being asked is that the default runs the other way here — unlisted first — and that is on purpose, not a limitation.

The literal answer

A link is shareable the moment it is published; “public” in the sense of anyone-with-the-link is the default. What is not the default is listed-and-indexed — crawlable, discoverable, in search results. That you turn on deliberately, because it is rarely what agent output wants.

Why the default is unlisted

Agent output is built from context, and context carries data that should not be indexed. So every viewer response ships noindex and a stripped referrer, and the URL is an unguessable token. The link works for the person you send it to and stays out of search for everyone else.

Widening reach, step by step

  1. Published: anyone with the link can open it; nothing indexes it.
  2. Want it findable: allow indexing on that link deliberately — a per-link choice, not a global switch.
  3. Want it restricted instead: add a password or an email-domain gate and it narrows, not widens.

The gate table, for when public is too much

You wantControl
A shared secretset_password
One organisationset_email_gate
A single viewBurn-after-read
A deadlineset_expiry

Public done on purpose

Landing pages, demos, things you want found — those genuinely want to be public and indexed, and nothing stops you. The design just refuses to make “indexed by the whole internet” the thing that happens when you forget to choose.

So is it public or private by default?

Shareable by default, indexed only on purpose. Anyone with the link can open it; no crawler is invited unless you decide to invite one.

Can I make it truly private, not just unlisted?

Yes — a password or an email-domain gate turns “anyone with the link” into “only these people”.

Keep reading