The five commands

npx intentmesh adopt          # writes .intent/config.json and a CI workflow
npx intentmesh file --id settlement-rails --kind initiative --title "..."
npx intentmesh promote intent/you/settlement-rails --by person/you --to public
npx intentmesh emit           # writes intent.fragment.json and the public projection
npx intentmesh merge ./fragments --tier public

When https://yourdomain/.well-known/intent.json answers, you are publishing. Filing is always private; promote is the one step that needs a named human, and that is the design, not a hoop. Then check yourself the way a stranger would:

npx intentmesh verify yourdomain.com
✓ published — https://yourdomain.com/.well-known/intent.json
  3 public item(s), 2 naming wants the network can match

As a library

import { file, promote, merge, view, validateFragment } from 'intentmesh'

const item = file({ id: 'intent/you/thing', kind: 'task', title: 'A thing' /* … */ })
// item.visibility === 'private' — there is no argument that changes this

const { intents, problems } = merge([yours, theirs])
const publicOnly = view(intents, 'public')

npm i intentmesh — zero runtime dependencies, and governance commits to keeping it that way.

Without a toolchain

vendor-intentmesh.mjs is the whole format as one dependency-free file: copy it into your repository, run it with bare node, get a valid fragment. It exists because the organisations most likely to adopt a publication format are often the ones least interested in installing a toolchain to do it.

What adopting signs you up for

Nothing. There is no registration, no API key, no callback to a server we run, and no field whose value only we can issue — a valid fragment can be produced, published, read and merged by somebody who has never heard of this company. If you want the world to know you publish, the registry is a pull request; if you don't, your file still works.