SRE Security Model
The Smyth Runtime Environment (SRE) is designed with security at its core. Every operation, agent execution, resource access, or collaboration, is protected by a security model that includes zero-trust access control, vault-backed secret management, and real-time role enforcement.
Identity Enforcement with Candidates and ACLs
SRE enforces access at the identity level using the Candidate and ACL model:
- A
Candidateis any identity in the system -> an agent, user, or even a specific process. - An Access Control List (ACL) defines what actions each candidate can perform.
Every time a candidate attempts to access a resource, such as storage or an LLM, SRE checks the ACL for explicit permission.
const candidate = AccessCandidate.agent(agentId); const storage = ConnectorService.getStorageConnector().user(candidate); await storage.write('results.json', data);
You can manage these permissions through Roles and Permissions, and they are enforced consistently across Spaces and your Organization.
Vault-Based Secret Management
Sensitive data, like API keys, credentials, and tokens, is never hardcoded or exposed. Instead, SRE uses the Vault subsystem for secure, encrypted secret storage.
- Vaults encrypt all data at rest and in transit.
- You can use built-in vaults or external providers, including AWS Secrets Manager and HashiCorp Vault.
- Role-based access ensures only approved agents or users can access specific secrets.
- Every secret access is logged for compliance and traceability.
To manage secrets, visit the Vault dashboard.
Authentication and Identity Providers
SRE supports both built-in authentication and integration with external enterprise systems:
- Connect with enterprise SSO or LDAP.
- Add custom account connectors to fit your organization’s authentication workflows.
- Use token-based login for programmatic or API access.
You can manage users and authentication settings from the Organization Management panel.
Role-Based Access Control (RBAC)
Role-based access control is central to SRE. Roles define who can do what... from the agent runtime to your admin dashboard, and are enforced in real time.
- Organization-wide roles: e.g., Super Admin
- Space-specific roles: e.g., Admin, Editor, Viewer
If a candidate does not have the correct permissions, the action is denied before execution.
To adjust or review roles, see the Roles and Permissions guide.
Enterprise Security Features
SRE is built to support production and regulated environments:
- TLS/SSL encryption across all internal and external communication.
- Audit logs for every critical operation (agent runs, data access, vault usage).
- Vault secrets can be rotated and revoked automatically.
- Least privilege policies since unknown or unconfigured actions are denied by default.
To learn more about deploying SRE in enterprise settings, see Enterprise Deployment.
How SRE Security Connects
SRE security spans multiple subsystems. Here’s where to manage each area:
| Security Area | Where It’s Managed |
|---|---|
| Identity & ACLs | Roles and Permissions |
| Credential Storage | Vault |
| User Roles & Login | Organization Management |
| Agent Environments | Spaces |
| Advanced Deployment | Enterprise Deployment |