Skip to content
Squad AI

Security & Governance

Squad is built for high-stakes environments where AI actions require oversight, traceability, and strict access controls. Security is layered throughout the platform.

Authentication

Squad integrates with your organisation’s identity provider via standard protocols.

  • Protocol: OAuth2/OIDC (industry standard)
  • SSO: SAML and OIDC integration with your existing identity provider
  • Token format: JWT with configurable expiry
  • Scope: All API endpoints require authentication unless explicitly whitelisted (health checks, OAuth config)

Authorisation

Squad uses two roles — user and admin — mapped from your identity provider via Keycloak client roles. Both roles carry full read and write access to the platform API (querying, reviewing, workflow execution). The admin role adds user management and system-wide configuration privileges. For the full permission reference, see Administration & Access Control.

A second layer of access control operates inside the AIM agent itself: non-admin users receive a restricted tool surface that excludes graph-mutation and external-search tools, regardless of their API permissions. This means a standard user can approve or reject a review through the API, but the agent will not execute a destructive Cypher write on their behalf.

AI Safety Controls

Security Review

Deterministic (not LLM-based) security checks are applied to every generated query. Dangerous operations are blocked regardless of user role. This enforcement is consistent and cannot be bypassed.

Risk-Aware Routing

Queries can carry a risk level. High-risk queries without a strong match to a proven, approved template are automatically declined: they never reach the execution stage.

Dynamic Tool Validation

When the system generates new tools, the code undergoes static analysis with restricted imports and execution constraints before being persisted.

Data Protection

Data at Rest

All data stores are protected with authentication. Encryption at rest is configured per your deployment model and compliance requirements.

Data in Transit

TLS encryption is applied to all external-facing connections. Internal service communication is isolated within the platform’s network boundary.

Audit Trail

Every interaction with the platform is fully traceable:

  • Correlation ID: Unique identifier per execution, flows through all logs and events
  • Structured logging: Session, user, and event type metadata
  • Persistent records: Every interaction is recorded for compliance and investigation
  • Real-time events: Include correlation IDs for client-side trace reconstruction

The audit trail is immutable: events are append-only with no deletion mechanism.

Compliance

Squad’s security controls are designed to support compliance with common regulatory frameworks. During onboarding, our team works with you to configure the platform to meet your specific compliance requirements.

For questions about specific compliance certifications or security assessments, contact our team.

Responsible Disclosure

If you discover a security vulnerability, please report it responsibly: