Role-based access control
Built-in roles
Union.ai provides three user-assignable roles with progressively broader permissions.
| Role | Capabilities |
|---|---|
| Admin | Full access: manage users, clusters, secrets, projects, all runs |
| Contributor | Create/abort runs, deploy tasks, manage secrets within assigned projects |
| Viewer | Read-only: runs, actions, logs, reports |
Additional internal system roles exist for platform operations but are not user-visible or user-assignable.
Custom policies
Roles (built-in or custom) define sets of permitted actions; custom policies then bind those roles to resources — scoped org-wide, by domain, or by project+domain — using composable YAML bindings. Both are managed with the flyte CLI (flyte create role / flyte create policy). This allows organizations to define fine-grained access policies that match their team structure and security requirements.
Enforcement
Every request is authenticated and authorized against the user’s role assignments before any data access occurs. Enforcement happens at two points: the control plane API gates orchestration metadata operations (run lifecycle, task definitions, RBAC management, project/domain queries), and the data plane’s dataproxy service gates every customer-data request (signed URLs, structured I/O retrieval, log fetching, auxiliary UI proxying, secret writes) after the Envoy router inside the customer’s cluster has authenticated it. For each decision, dataproxy asks the control plane’s authorizer, sending the caller’s identity and the requested action, not customer data. Both enforcement points evaluate the same Union.ai identity against the same RBAC policy.
Least privilege
Union.ai enforces least privilege across all components. IAM roles on the data plane are scoped to minimum required permissions. Each data plane has two IAM roles: an admin role for platform services and a user role for task pods. IAM roles are bound via cloud-native workload identity federation, eliminating static credentials entirely. Presigned URLs grant single-object, operation-specific, time-limited access. Service accounts receive only the permissions needed for their specific function.
Verification
RBAC enforcement
Reviewer focus: Confirm that each role enforces the expected permissions and that custom policies correctly scope access to specific projects and domains.
How to verify:
-
Create three test users with Admin, Contributor, and Viewer roles.
-
Log in as Viewer and confirm restricted operations are denied:
flyte run ... # Expect denied flyte create secret ... # Expect denied flyte get run # Expect success -
Log in as Contributor scoped to project A:
flyte run --project B ... # Expect denied flyte run --project A ... # Expect success -
Create a custom policy scoping a user to project X, development domain only. Attempt to access the production domain. Expect denied.
-
Display all active policy bindings:
flyte -of json get policy -
For Union.ai employee access, get the subjects of the staff identities in your tenant from Union.ai (
flyte get memberandflyte get userdon’t list support staff). Look up the policies assigned to each identity withflyte get assignment --user-subject <subject>, inspect those policies withflyte get policy <name>, then inspect the roles those policies bind withflyte get role <name>. The role carries the action list, so the last step is the one that shows the effective access. See Human access controls.
Steps 1–5 are fully self-service. Step 6 also needs the staff subjects from Union.ai, and assignment output shows only customer-visible policies, not internal system policies.