Identity#

The ATP identity system has a number of requirements:

• ID provision. Users should be able to create global IDs which are stable across services. These IDs should rarely change to ensure that links to their content are stable.
• Public key distribution. Distributed systems rely on cryptography to prove the authenticity of data and provide end-to-end privacy. The identity system must publish their public keys with strong security.
• Key rotation. Users must be able to rotate their key material without disrupting their identity.
• Service discovery. To interact with users, applications must be able to discover the services in use by a given user.
• Usability. Users should have human-readable and memorable names.
• Portability. Identities should be portable across services. Changing a provider should not cause a user to lose their identity, social graph, or content.

Adopting this system should give applications the tools for end-to-end encryption, signed user data, service sign in, and general interoperation.

Identifiers#

We use two interrelated forms of identifiers: the handle and the DID. Handles are DNS names while DIDs are an emerging W3C standard which act as secure & stable IDs.

The following are all valid user identifiers:

@alice.host.com
at://alice.host.com
at://did:plc:bv6ggog3tya2z3vxsub7hnal

The relationship between them can be visualized as:

┌──────────────────┐                 ┌───────────────┐ 
│ DNS name         ├──resolves to──→ │ DID           │
│ (alice.host.com) │                 │ (did:plc:...) │
└──────────────────┘                 └─────┬─────────┘
       ↑                                   │
       │                               resolves to
       │                                   │
       │                                   ↓
       │                            ┌───────────────┐ 
       └───────────references───────┤ DID Document  │
                                    │ {"id":"..."}  │
                                    └───────────────┘

The DNS handle is a user-facing identifier — it should be shown in UIs and promoted as a way to find users. Applications resolve handles to DIDs and then use the DID as the stable canonical identifier. The DID can then be securely resolved to a DID document which includes public keys and user services.

DID Methods#

The DID standard supports custom "methods" of publishing and resolving DIDs to the DID Document. A variety of existing methods have been published so we must establish criteria for inclusion in this proposal:

• Strong consistency. For a given DID, a resolution query should produce only one valid document at any time. (In some networks, this may be subject to probabilistic transaction finality.)
• High availability. Resolution queries must succeed reliably.
• Online API. Clients must be able to publish new DID documents through a standard API.
• Secure. The network must protect against attacks from its operators, a MITM, and other users.
• Low cost. Creating and updating DID documents must be affordable to services and users.
• Key rotation. Users must be able to rotate keypairs without losing their identity.
• Decentralized governance. The network should not be governed by a single stakeholder; it must be an open network or a consortium of providers.

At present, none of the DID methods meet our standards fully. Therefore we have chosen to support did-web and a temporary method we've created called did-placeholder. We expect this situation to evolve as new solutions emerge.

Handle Resolution#

Handles in ATP are domain names which resolve to a DID, which in turn resolves to a DID Document containing the user's signing pubkey and hosting service.

Handle resolution uses the com.atproto.handle.resolve XRPC method. The method call should be sent to the server identified by the handle, and the handle should be passed as a parameter.

Here is the algorithm in pseudo-typescript:

async function resolveHandle(handle: string) {
  const origin = `https://${handle}`
  const res = await xrpc(origin, 'com.atproto.handle.resolve', {handle})
  assert(typeof res?.did === 'string' && res.did.startsWith('did:'))
  return res.did
}

Example: Hosting service#

Consider a scenario where a hosting service is using PLC and is providing the handle for the user as a subdomain:

• The handle: alice.pds.com
• The DID: did:plc:12345
• The hosting service: https://pds.com

At first, all we know is alice.pds.com, so we call com.atproto.handle.resolve() on alice.pds.com. This tells us the DID.

await xrpc.service('https://alice.pds.com').com.atproto.handle.resolve() // => {did: 'did:plc:12345'}

Next we call the PLC resolution method on the returned DID so that we can learn the hosting service's endpoint and the user's key material.

await didPlc.resolve('did:plc:12345') /* => {
  id: 'did:plc:12345',
  alsoKnownAs: `https://alice.pds.com`,
  verificationMethod: [...],
  service: [{serviceEndpoint: 'https://pds.com', ...}]
}*/

We can now communicate with https://pds.com to access Alice's data.

Example: Hosting service w/separate domain name#

Suppose we have the same scenario as before, except the user has supplied their own domain name:

• The handle: alice.com (this differs from before)
• The DID: did:plc:12345
• The hosting service: https://pds.com

We call com.atproto.handle.resolve() on alice.com to get the DID.

await xrpc.service('https://alice.com').com.atproto.handle.resolve() // => {did: 'did:plc:12345'}

Then we resolve the DID as before:

await didPlc.resolve('did:plc:12345') /* => {
  id: 'did:plc:12345',
  alsoKnownAs: `https://alice.com`,
  verificationMethod: [...],
  service: [{serviceEndpoint: 'https://pds.com', ...}]
}*/

We can now communicate with https://pds.com to access Alice's data. The https://alice.com endpoint only serves to handle the com.atproto.handle.resolve() call. The actual userdata lives on pds.com.

Example: Self-hosted#

Let's consider a self-hosting scenario. If using did:plc, it would look something like:

• The handle: alice.com
• The DID: did:plc:12345
• The hosting service: https://alice.com

However, if the self-hoster is confident they will retain ownership of the domain name, they can use did:web instead of did:plc:

• The handle: alice.com
• The DID: did:web:alice.com
• The hosting service: https://alice.com

We call com.atproto.handle.resolve() on alice.com to get the DID.

await xrpc.service('https://alice.com').com.atproto.handle.resolve() // => {did: 'did:web:alice.com'}

We then resolve using did:web:

await didWeb.resolve('did:web:alice.com') /* => {
  id: 'did:web:alice.com',
  alsoKnownAs: `https://alice.com`,
  verificationMethod: [...],
  service: [{serviceEndpoint: 'https://alice.com', ...}]
}*/