---

title: NIP-AD web addresses audience: user

NIP-AD web addresses

A relay or site URL works in a browser and resolves to a Nostr event in clients that support NIP-AD. bind.ws follows proposal 2406, AD.md at b82a9bf66ec9757149b5aa3b3cd36190dd4e6fed, checked 2026-09-04. The proposed text labels itself final and optional, but the pull request is still a proposal. Its wire format may change.

Resolve a URL

A supporting client takes the URL's pathname and requests https://<host>/.well-known/nostr.json?path=<encoded-pathname>. Encode the pathname as a query parameter, preserving any escapes already in the URL: /a/hello%20world becomes path=%2Fa%2Fhello%2520world.

For https://alice.bind.ws/a/why-relays, the response has this shape:

{
  "/a/why-relays": {
    "filter": {
      "kinds": [30023],
      "authors": ["<owner-pubkey>"],
      "#d": ["why-relays"],
      "limit": 1
    },
    "relays": ["wss://alice.bind.ws"]
  }
}

The client queries the named relay with that filter. A browser opening the original URL continues to receive the existing page. Notes and articles also have an "Open in a nostr client" link.

Browser URL	Nostr counterpart
Relay /	Relay-signed kind 39000, with the relay identity and group identifier
Relay /e/<id>	That exact note or article event ID, including its version
Relay /a/<d>	Owner's current kind 30023 article with that identifier
Relay /a/<author>/<d>	That author's current article; hex, npub and nprofile authors follow the browser page parser
Site /, a mapped file or directory	Current kind 15128 or 35128 manifest, or exact kind 5128 snapshot

All filters include limit: 1 and explicit relay hints. Address filters include the full author, kind and, for addressable events, #d. Event ID filters use the full ID. Site hints name the relay holding the manifest, not the site hostname. Custom domains follow their selected relay or site; local development retains ws:// and its port.

Visibility and errors

Discovery applies the target's existing access rules:

Target	Rule
Relay group	Current read rule; private metadata requires an admitted NIP-98 identity
Note or article	Open reads and the pages feature; signing does not open private pages
Site manifest	sites feature, site authentication and the current read rule
Removed, expired or moderation-hidden event	No mapping
Unclaimed or expired relay	No mapping

Successful lookups return only the requested path. Unknown paths return {}; discovery does not return the member directory or enumerate paths. Site paths must appear in the live manifest, directly or as a directory's index.html. A /404.html fallback does not give a missing path a mapping. Discovery reads manifest metadata without fetching or verifying file bytes; the normal site request still verifies and serves the file.

Private group and site requests return 401 without authentication or 403 for a proved identity the read rule excludes. Site visitors retain the existing sign-in flow and host-only cookie; API clients can sign the exact discovery URL with NIP-98. Site-origin discovery never exposes the hosting relay's NIP-05 names, console, WebSocket, event pages or management methods. Blocked IP addresses remain blocked.

GET returns JSON, HEAD returns the same status and headers without a body, and OPTIONS supports discovery preflight without resolving a path. Discovery responses use CORS and Cache-Control: private, no-store so policy changes are checked each time. Unsupported methods return 405. Invalid GET or HEAD paths return 400: only one absolute pathname of at most 4096 characters is accepted, without literal queries, fragments, whitespace, control characters, backslashes or dot segments. Percent-encoded spaces and other article identifier characters work when the pathname is correctly query-encoded. Full URLs are not accepted and discovery never fetches an external URL.

Names and capabilities

The existing NIP-05 endpoint keeps its behavior. With name, including a request that also has path, it answers the name lookup. Without either parameter it answers the permitted member directory. Turning off names disables those NIP-05 responses; path discovery remains available. Turning off pages or sites removes their respective mappings.

This implementation leaves the NIP-11 supported_nips representation unchanged. Lettered capability advertisement is coordinated separately with NIP-9a and NIP-5A compatibility work.