Security policy

Security · Source: SECURITY.md

Security Policy

cubby-network is an open-source NetOps control plane that, in production, holds credentials for real network devices and can change live network state. Vulnerability reports are taken seriously.

Reporting a vulnerability

Do not open a public GitHub issue for a security report. Instead:

  1. Email the maintainers with [cubby-network security] in the subject line.
  2. Include reproduction steps, the version / commit you tested, and the impact you believe the issue has.
  3. We will acknowledge within 72 hours and aim to provide a remediation plan within 14 days for critical issues, 30 days for high severity.

When in doubt, report it. We would much rather see a non-issue than miss a real one.

Supported versions

Until the project reaches v1.0.0 we patch only the main branch. Once we cut a stable release, we will support the latest minor for security fixes.

Threat model summary

The harness is designed against the following classes of attack:

Things we explicitly do not protect against (yet)

Signing-key custody

The harness uses two distinct keys: one for evidence bundles, one for change approvals. They are intentionally separate so rotating or compromising one does not invalidate or forge the other.

Default custody (dev / demo): both keys are file-backed HMAC-SHA256 secrets auto-generated on first use under var/keys/ (0600 permissions, 32 random bytes). Deterministic across processes on the same host. Suitable for local development, CI, and single-operator demos — not production.

Production custody (recommended):

The evidence_require_configured_key flag applies to both signers; one switch flips the whole harness into strict-keys mode.

Hardening checklist for production