Skip to content

Security: LaurierCS/locah-ai

Security

SECURITY.md

Security & Privacy Disclosure

LOCAH.ai handles student questions. If you find a vulnerability — especially anything that could expose a question to another user, leak identifying information, or bypass the redaction layer — please report it privately rather than opening a public issue.

Contact: <security contact email — TBD>

Please include what you found, how to reproduce it, and what you think the impact is. We will acknowledge within 5 days.

What we consider in scope

  • Any path by which personally identifying information reaches the model provider (violates INV-2).
  • Any path by which a question is persisted with identity (violates INV-3).
  • Any path by which the safety gate can suppress or de-prioritize a message (violates INV-5).
  • Aggregate reports that permit re-identification (violates INV-6).
  • Standard web vulnerabilities: injection, SSRF via the crawler, auth bypass on admin endpoints.

See docs/SDD.md §3 for the full invariant list.

There aren't any published security advisories