# Enterprise

Enterprise features for business organizations

SteelEngine Enterprise provides advanced features for organizations with enhanced security, compliance, and management requirements.

***

## Access Control

Permission groups collect people and can receive PBAC role assignments at organization or workspace scope. Enterprise feature restrictions can additionally limit which models, blocks, and platform features group members can use.

### Features

* **Model Providers** - Restrict which AI providers group members can use (OpenAI, Anthropic, Google, etc.)
* **Blocks** - Control which workflow blocks and integrations are available
* **Platform** - Hide workspace pages and settings tabs (Knowledge Base, Tables, Files, API Keys, Environment), the workflow editor's **Assistant** tab, and trace spans in logs. Hide Deploy tabs (API, MCP, A2A, Hosted Chat). Disable MCP tools, custom tools, Skills, and invitations

### Setup

1. Navigate to **Organization Settings → Access → Groups**.
2. Create a permission group and add members.
3. Assign one or more seeded or custom roles to the group at organization or workspace scope.
4. Configure any restrictions on the group's **Model Providers**, **Blocks**, and **Platform** tabs.

Not belonging to a group does not imply full access. PBAC roles and assignments always determine resource authority. Enterprise group restrictions are an additional product-feature filter enforced in the UI and at execution time.

***

## Single Sign-On (SSO)

Enterprise authentication with SAML 2.0 and OIDC support for centralized identity management.

### Supported Providers

* Okta
* Azure AD / Entra ID
* Google Workspace
* OneLogin
* Any SAML 2.0 or OIDC provider

### Setup

SteelEngine sets up SSO for your organization. Contact your SteelEngine account team to connect your identity provider or change its settings. Have these ready:

1. The sign-in protocol (**OIDC** or **SAML**)
2. The email domain to route to your identity provider, and its issuer URL
3. The protocol credentials. OIDC needs a client ID, client secret, and scopes. SAML needs an entry point URL and certificate

Once the connection is active, **Organization Settings → SSO** shows the provider, its email domain, and the callback URL to register with your identity provider. Only SteelEngine staff can add, change, or remove an SSO provider.

Once SSO is enabled, team members authenticate through your identity provider instead of email/password.

***

## Self-Hosted Configuration

For self-hosted deployments, enterprise features can be enabled via environment variables without requiring billing.

For Azure DevOps deployment applications, configure [Entra workload identities](/self-hosting/entra-workload-identities) separately from human SSO. The installation accepts a configured resource API and tenant; organization administrators register applications and assign their local workspace actions in **Workload identities**. This path does not require the SSO/SCIM connection or a human service account.

### Environment Variables

| Variable                                                         | Description                                         |
| ---------------------------------------------------------------- | --------------------------------------------------- |
| `ORGANIZATIONS_ENABLED`, `NEXT_PUBLIC_ORGANIZATIONS_ENABLED`     | Enable team/organization management                 |
| `ACCESS_CONTROL_ENABLED`, `NEXT_PUBLIC_ACCESS_CONTROL_ENABLED`   | Permission groups for access restrictions           |
| `SSO_ENABLED`, `NEXT_PUBLIC_SSO_ENABLED`                         | Single Sign-On with SAML/OIDC                       |
| `CREDENTIAL_SETS_ENABLED`, `NEXT_PUBLIC_CREDENTIAL_SETS_ENABLED` | Polling Groups for email triggers                   |
| `DISABLE_INVITATIONS`, `NEXT_PUBLIC_DISABLE_INVITATIONS`         | Globally disable workspace/organization invitations |

### Organization Management

When billing is disabled, use the Admin API to manage organizations.

The Admin API authenticates with the `x-admin-key` header (the `ADMIN_API_KEY` environment variable). It also covers users, workspaces (import/export), workflows (import/export, deploy, version activation), subscriptions, credits, audit logs, and access control. It is an operator surface, and this page does not document it further.

The Admin API addresses orgs and workspaces by their **internal UUIDs** (`{orgId}`, `{workspaceId}`), not by the URL slugs. UUIDs are stable across renames; slugs are not. The user-facing dashboard URLs (`/{orgSlug}/{workspaceSlug}/...`) and the Admin API are deliberately separate — see [Organizations and Workspaces](/organizations) for the full slug-vs-UUID model.

```bash
# Create an organization
curl -X POST https://your-instance/api/v1/admin/organizations \
  -H "x-admin-key: YOUR_ADMIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "My Organization", "ownerId": "user-id-here"}'

# Add a member
curl -X POST https://your-instance/api/v1/admin/organizations/{orgId}/members \
  -H "x-admin-key: YOUR_ADMIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"userId": "user-id-here", "role": "admin"}'
```

### Workspace Members

When invitations are disabled, use the Admin API to manage workspace memberships directly:

```bash
# Add a user to a workspace
curl -X POST https://your-instance/api/v1/admin/workspaces/{workspaceId}/members \
  -H "x-admin-key: YOUR_ADMIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"userId": "user-id-here", "roleKey": "workspace:write"}'

# Remove a user from a workspace
curl -X DELETE "https://your-instance/api/v1/admin/workspaces/{workspaceId}/members?userId=user-id-here" \
  -H "x-admin-key: YOUR_ADMIN_API_KEY"
```

### Notes

* Enabling `ACCESS_CONTROL_ENABLED` automatically enables organizations, as group access management requires organization membership.
* When `DISABLE_INVITATIONS` is set, users cannot send invitations. Sign-up is invitation-only, so new accounts then come only from SSO or the Admin API (`POST /api/v1/admin/users`). Use the Admin API to manage workspace and organization memberships too.

## Common Questions

**What are the minimum requirements to self-host SteelEngine?**

The Docker Compose production setup (docker/selfhost/compose.yml) includes the SteelEngine application (8 GB memory limit), the Temporal Worker (4 GB memory limit), a realtime collaboration server (1 GB memory limit), a one-shot migrations service, and a PostgreSQL database with pgvector. A machine with at least 16 GB of RAM and 4 CPU cores is recommended. You will also need Docker and Docker Compose installed.

**Can I run SteelEngine completely offline with local AI models?**

Yes. SteelEngine supports Ollama and VLLM for running local AI models. A separate Docker Compose file (docker/selfhost/compose.ollama.yml) is available for deploying with Ollama. This lets you run workflows without any external API calls, keeping all data on your infrastructure.

**How does data privacy work with self-hosted deployments?**

Workflow definitions, execution logs, and user data use your installation's configured storage. PostgreSQL-backed credentials are encrypted. With optional Azure Key Vault storage, SteelEngine resolves eligible secrets from your approved vault and stores references in PostgreSQL. External AI providers receive the inputs and credentials required for their calls.

**Do I need a paid license to self-host SteelEngine?**

Self-hosting SteelEngine requires an Enterprise subscription. Contact sales@steelengine.com for pricing and deployment options.

**Which SSO providers are supported?**

SteelEngine supports SAML 2.0 and OIDC protocols, which means it works with virtually any enterprise identity provider including Okta, Azure AD (Entra ID), Google Workspace, and OneLogin. SteelEngine sets it up for your organization, and Organization Settings > SSO shows the connection.

**How do I manage users when invitations are disabled?**

Use the Admin API with your admin API key. Workspace membership requests use a roleKey such as workspace:write or a custom role key; the tier-style permissions field is rejected. Organization membership and workspace role assignment are separate operations.

**Can I scale SteelEngine horizontally for high availability?**

The Docker Compose setup is designed for single-node deployments. For production scaling, you can deploy on Kubernetes with multiple application replicas behind a load balancer. The database can be scaled independently using managed PostgreSQL services. Redis can be configured for session and cache management across multiple instances.

**How do access control permission groups work?**

A group can receive PBAC role assignments, so members inherit the group's resource permissions at the assignment scope. Enterprise restrictions can also narrow available model providers, workflow blocks, and platform features. Both layers are enforced; a feature restriction never grants a missing PBAC permission.
