intrnl.cloud
Identity and security

Governance and ownership

Roles, risk classes, capability floors, approval authority, and organization-owned app lifecycles.
v1 design baseline. This page specifies intended behavior. Delivery and validation are tracked in the implementation plan; it is not a claim that the platform is already implemented.

Organization roles

RoleTypical authority
ViewerView permitted apps and deployments.
CreatorCreate apps, edit managed source, build, preview, request deployment.
DeveloperGit/API access, advanced diagnostics, source ownership, technical maintenance.
DeployerDeploy within delegated risk/capability limits.
ApproverApprove designated deployment classes.
Security AdminIdentity, IP lists, policies, secrets, integrations, capability approval.
Org AdminOrganization-wide administration and role assignment.

Permissions should remain granular even if the UI initially presents role bundles.

Risk classes

Sandbox

  • Synthetic data
  • Creator/reviewer access only
  • No production integrations
  • No secrets
  • Short retention
  • Self-service preview

Standard internal

  • Entra SSO
  • Normal internal workflow data
  • SQLite and KV
  • Controlled egress
  • Organization/group access
  • No regulated data

Sensitive

  • Privileged internal data
  • Secrets or external integrations
  • Broad departmental use
  • Expanded access policy
  • Stronger approvals and logging

Regulated

  • Regulated data
  • Integrations subject to compliance requirements
  • Contractual/compliance controls
  • Approved AI configuration
  • Dedicated policy and retention
  • Formal review

The v1 organization pilot enables only Sandbox and Standard Internal.

Capability-driven minimum risk

The creator does not get to misclassify an app as low risk.

Examples:

Requests only DB + KV + Entra SSO
→ Standard may be sufficient

Requests generic raw secret
→ at least Sensitive

Requests private-network connector
→ at least Sensitive

Requests regulated-data capability
→ Regulated

Requests public exposure
→ security approval and organization permission

An administrator may raise the risk class but may not lower it beneath the computed capability floor.

Approval matrix

Pilot policy:

ChangeApproval
Sandbox previewCreator
Standard internal production deploymentAuthorized human approver
Destructive migrationAuthorized approver
New external integrationSecurity Admin
Access expansionSecurity Admin
Public exposureDisabled
Private networkDisabled
Regulated-data capabilityDisabled
Organization policySecurity Admin/Org Admin
AI self-approvalNever

Application ownership

Every production app has:

Organization owner
Business owner
Technical owner/maintainer
Risk class
Data classification
Last review date
Next review date

When the creator leaves:

  • App remains organization-owned.
  • Direct credentials are revoked.
  • Business/technical ownership is reassigned or flagged.
  • Production is not automatically deleted.
  • Administrators receive an orphaned-app alert.

Lifecycle

Draft
Preview
Production
Suspended
Archived
Pending Deletion
Purged

Inactivity policy examples:

No production traffic for 180 days
→ flag for review

No production traffic for 365 days
→ archive candidate

Deletion initially disables routes and starts a recovery window. Permanent purge follows organization retention policy.

Governance dashboard

Organization administrators should see:

Production apps by risk class
Apps with external egress
Apps with integrations
Apps with destructive migrations pending
Apps missing owners
Apps overdue for review
Apps with broad network access
Apps with policy exceptions
AI-created changes awaiting review
Backup/restore health
Quota consumption