VerifableStart a recordOpen app
Docs · Social publication

Publishing to the network

What composing a post through the wallet's connected AT Protocol account actually does, from the character limit to the disclosure preview to the audit trail. See connecting an account first if the wallet has no connected account yet.

Composing

The composer writes ordinary Bluesky-compatible posts and replies — the same app.bsky.feed.post record type any AT Protocol client writes, with the standard 300-character limit. There is nothing proprietary in the record itself.

Attaching a passport

A post can attach the public passport page as a link card. The card is built from the same public page anyone can open — nothing restricted is summarised into it. Because the attachment is an ordinary external link embed, a native Bluesky client sees a complete, useful post, with no Verifable-specific rendering required.

The disclosure preview

Before a passport-linked post can publish, the disclosure preview shows two lists side by side: the public fields the post includes, and the restricted or personal fields it excludes. An explicit confirmation is required whenever any of the following is true:

  • the post names a location more precisely than the passport's public granularity;
  • the post contains personal information;
  • an agent prepared the post;
  • the post discloses a professional role; or
  • the wording implies a finance-related commitment.

Cancelling at the preview publishes nothing.

Publication context

The current composer publishes only in the personal context. The model contains an organisational-context shape, but it is deliberately not offered until a real Wallet authority read supplies the organisation, mandate and policy. The author DID of the connected AT Protocol account never becomes a Wallet or organisation DID by implication.

Evidence language

Every statement Verifable publishes or imports carries one of five provenance labels:

  • Verified from passport — read directly from a signed passport field.
  • Derived by Veritas — computed from verified data, not itself a direct read.
  • Written by the author — the person's own words, unverified by design.
  • Suggested by an agent — agent-prepared, pending human confirmation.
  • Still unverified — a claim with no supporting evidence yet.

Each label governs how subsequent operations may use the statement.