Single sign-on
A user signs in once and reaches every connected application. Authorization code flow with PKCE for applications, client credentials for services, rotating refresh tokens.
Product
You connect each application to OneiD once. Sign-in pages, second factors, sessions, roles and the record of who signed in are then the same for all of them.
What you get
The parts of OneiD that your users, your developers and your administrators work with.
A user signs in once and reaches every connected application. Authorization code flow with PKCE for applications, client credentials for services, rotating refresh tokens.
Authenticator-app codes (TOTP) with enrolment by QR code. Make the second factor optional, or mandatory for everyone or for chosen users.
One web console for applications, users, roles, scopes and sessions, with read-only and limited administrator roles.
Assign roles in OneiD or map them from directory groups. Applications read them from the token; they do not need to query a directory.
Sign-ins, failed attempts, password changes, client changes and key rotations are recorded with time and address.
Applications can sign the user out of OneiD. Administrators can revoke a user’s tokens at any time, and tokens are revoked when a password changes.
Signing keys are stored encrypted and survive restarts. A new key is published before it is used, so applications pick it up without interruption.
Rate limits on sign-in and token requests, strict browser security headers, HTTPS only, and no secrets in logs.
Admin console
Administrators work in a web console. Every action there is also available through the Admin API, under the same roles.
Register clients, their redirect addresses, scopes, allowed origins and token lifetimes. Generate, replace or disable a client secret.
Create users, assign roles, unlock accounts, reset the second factor, require a password change or a second factor.
Define roles and the API scopes your services accept. Built-in administrator roles cannot be renamed or deleted.
See who holds tokens for which application and revoke them.
Search sign-ins, failures, administrator changes and key rotations. Auditors get a read-only role of their own.
Rotate the keys that sign tokens, with an activation time so that the new key is published first.
Built-in roles separate who manages applications, who manages users and who only reads. The management roles also come in read-only versions. The Admin API accepts tokens only from administrators whose token carries the admin scope.
| Role | What it may do |
|---|---|
All | Full administrator. The only role that manages other administrators and privileged scopes. |
AllReadOnly | Sees everything, changes nothing. |
AuthorizationServerManager | Manages applications, scopes and keys. |
UserManager | Manages users and their roles. |
Auditer | Reads the audit log. |
Sign-in sources
OneiD can keep the accounts itself or check them against a system that you already run. Your applications do not see the difference: they always talk to OneiD and always receive the same kind of token. One OneiD installation uses one of these sources.
Users sign in with a user name and password that OneiD stores. Use this mode when you have no directory, or when you want OneiD to be the directory.
Users sign in with their directory user name and password. OneiD checks the password against the directory over LDAPS and never stores it.
OneiD sends the user to an external identity provider such as Okta, with the authorization code flow and PKCE. Your applications still talk only to OneiD and receive OneiD tokens.
A dedicated Microsoft Entra ID (Azure AD) sign-in source is planned. Talk to us if you need it.
Standards
OneiD speaks the protocols that sign-in libraries already know. It was tested with the OpenID Foundation’s conformance suite: the Config, Basic, Form Post and RP-Initiated Logout test plans for providers ran with no failed test. OneiD is not yet OpenID Certified.
Authorization code, client credentials and refresh token grants (RFC 6749)
Required for browser and mobile applications (RFC 7636)
ID tokens, userinfo, prompt and max_age, the claims parameter
Applications configure themselves from one address
Sign-out that starts in the application
Authorization responses by HTTP POST
RFC 7662 and RFC 7009
Signed access tokens that an API can check without a call to OneiD
Check this list before you plan an integration. OneiD does not offer:
Get OneiD
We can run OneiD for you, or you can run it in your own environment. Talk to us about the way that fits your organisation.
We run OneiD for you.
You run OneiD in your own cloud account or data centre.
Cannot find your answer? Ask us.
Not yet. OneiD was tested with the OpenID Foundation’s conformance suite: the Config, Basic, Form Post and RP-Initiated Logout test plans for providers ran with no failed test. We will say “certified” only when the OpenID Foundation has certified it.
Yes. OneiD can check passwords against LDAP or Active Directory, or send users to an OpenID Connect provider such as Okta. Groups become OneiD roles through a mapping you control. One OneiD installation uses one source of users.
A dedicated Microsoft Entra ID source is planned.
No. OneiD speaks OpenID Connect and OAuth 2.0. If an application only supports SAML, or you need SCIM provisioning, talk to us before you choose OneiD.
Authenticator-app codes (TOTP), optional or mandatory for everyone or for chosen users. SMS, passkeys and push notifications are not available.
Each customer has its own OneiD with its own database. When we host OneiD, you get your own OneiD address and data store. When you run it yourself, the data stays in your PostgreSQL database.
Yes. The documentation is published as Markdown and in llms.txt, and a connector specification lists exactly what an integration must do. Give your agent the specification and your client settings.
The demo application signs you in through OneiD and shows the tokens it receives. Talk to us about running OneiD for your organisation.