Skip to content

Security: Harikeshav-R/Understudy

Security

SECURITY.md

Security Policy

Supported Versions

Understudy is under active development. Security updates are applied to the latest release branch on main.

Version Supported
0.1.x ✅
< 0.1 ❌

Security Architecture & Invariants

Understudy is designed around a zero-trust execution model for AI-generated remediation actions:

  1. Closed Action Enum: The LLM cannot execute arbitrary shell commands or code in production. It may only synthesize typed plans conforming to contracts/models.py.
  2. Safety Kernel Verification (Z3): Every candidate plan that wins tournament scoring must be formally verified against system invariants before production actuation is permitted (Invariant K10).
  3. Triple-Layer Twin Egress Isolation:
    • Kubernetes NetworkPolicy denies egress from twin namespaces (except to the mirror gateway).
    • CoreDNS rewrite rules direct all outbound traffic to an egress stub.
    • Workload environment variables point sensitive external endpoints to mock sinks.
  4. Read-Only Production Boundary: Fleet controllers and twin workers have read-only access to ust-prod. Only the orchestrator's actuate node holds actuation permissions.
  5. No Secret Ingestion: Secrets and credentials are never stored in run records, playbooks, logs, or commit history.

Reporting a Vulnerability

We take the security of Understudy seriously. If you discover a security vulnerability or an isolation bypass, please report it privately rather than opening a public issue.

How to Report

  1. Open a GitHub Security Advisory under the repository's "Security" tab, or
  2. Email the maintainers at security@understudy.dev with:
    • A description of the vulnerability and its potential impact.
    • Step-by-step reproduction instructions or a proof-of-concept.
    • Any proposed remediation or mitigation.

Response Commitment

  • Initial Acknowledgement: Within 48 hours.
  • Triage & Assessment: Within 5 business days.
  • Resolution & Disclosure: Coordinated release and public advisory once a fix is verified.

There aren't any published security advisories