Skip to content

Security: insightitsGit/PrismGuard

Security

SECURITY.md

Security Policy

Supported versions

Version Supported
0.1.x (latest on PyPI) Yes
Older than latest patch on PyPI Best effort — please upgrade

Reporting a vulnerability

Please do not open a public GitHub issue for security vulnerabilities.

Report privately to:

Include:

  1. Description of the issue and impact
  2. Steps to reproduce (PoC if available)
  3. Affected version / commit
  4. Whether the issue is already public elsewhere

We aim to acknowledge reports within 5 business days and to provide a remediation plan or status update within 14 business days.

Scope

In scope examples:

  • Remote code execution or arbitrary file write via the library / CLI / HTTP service
  • Authentication / authorization bypass of license or HTTP controls
  • Prompt-injection bypass of PrismGuard itself that is reproducible and high impact (with clear repro)
  • Secrets accidentally shipped in the repository or package

Out of scope / lower priority examples:

  • Theoretical attacks without a practical repro
  • Issues that require already-compromised host credentials
  • Social engineering of maintainers
  • Vulnerabilities only in third-party dependencies with no PrismGuard-specific exploit path (please report upstream; we will upgrade when fixed)

Safe harbor

We will not pursue legal action against researchers who:

  • Make a good-faith effort to avoid privacy violations and service disruption
  • Do not access data that is not their own
  • Report findings promptly and privately
  • Do not exploit the issue beyond what is needed to demonstrate it

Hardening notes for operators

  • Keep PRISMGUARD_DEV_UNRESTRICTED unset in production
  • Use a production license public key before issuing paid licenses (see docs/enterprise-product-model.md)
  • Prefer network isolation for prismguard-serve and authenticate at your edge
  • Treat model artifacts and seed corpora as trusted inputs; verify downloads (prismguard-model download)

There aren't any published security advisories