Sign in through your IdP.
Okta and any OIDC-compatible provider plug straight in.
Sign-in through your identity provider, isolated per tenant, revocable in one action.
“We’ll add SSO later” is where agent pilots go to die in security review.
Your provider owns who exists; Fabriq resolves what they may do — org, then team, then member, and anything not granted is denied.
session open · grants inherited from the groupSigned in through your provider, in the Engineering group. The agent acts as Jo, with Jo’s own reach.
The grant resolves org, then team, then member — and Engineering has this one.
A named human approver has to say yes first. The role is unchanged while the request waits.
session cut · no identity to act forCutting the session ends it there and then — it does not run on until a token expires.
The group membership is gone, so the grant it carried is gone with it. Nothing to inherit.
An identity that no longer exists cannot be approved into a bigger one. The request dies with it.
Illustrative · your provider owns who exists; nothing here keeps a second copy of that answer.
Okta and any OIDC-compatible provider plug straight in.
Standard token exchange behind sign-in — nothing homegrown to review.
Cut a session and it dies immediately — not at token expiry.
Platform admin sits behind its own auth, TOTP, and recovery codes.
Organizations, teams, and invitations are first-class from the start.
Grants resolve org → team → member, default-deny, with per-user overrides.
SIGN IN THROUGH YOUR IDP — REVOKE A SESSION AND IT DIES NOW, NOT AT EXPIRY
Explore the context attached to every request.
workplace-agent has a defined purpose and a bounded set of tools.
The request acts for active employee, using that person’s permissions.
The policy applies to the requested action within Identity provider, before the tool executes.
Okta and OIDC-compatible providers connect directly — people sign in the way they already do, and agents operate under those identities.
Yes — platform administration is a separately hardened plane with its own authentication, TOTP, and recovery codes.
Yes — organizations contain teams with invitations and grants at each level, and permissions resolve org → team → member on a default-deny basis.
Immediately — session revocation takes effect now, not at the next token expiry. Disabling an agent covers the non-human side just as fast.
Yes — every agent is registered with an owner and scopes, and every action carries both the agent's identity and the signed-in user it acted for.