Skip to content

Security: cgrtml/reasongate

Security

SECURITY.md

Security Policy

ReasonGate is a security tool; we hold it to a security tool's standard.

Supported versions

The 0.1.x line receives security fixes. Pre-1.0 the API may change; pin a version in production.

Reporting a vulnerability

Please report suspected vulnerabilities privately — do not open a public issue.

  • Email: cagritemelusa@gmail.com with subject SECURITY: reasongate.
  • Include a description, affected version, and a minimal reproduction if possible.

We aim to acknowledge within 5 business days and to coordinate a fix and disclosure timeline with you. Please give us reasonable time to remediate before public disclosure.

Scope notes

  • The gate is one layer, not a guarantee. ReasonGate reduces prompt-injection / jailbreak risk; it does not eliminate it. Run it with the model's own safety training behind it. Recall is never 100%.
  • Adversarial input hardening. The core parses untrusted text; oversized or pathological inputs should be bounded by the caller (see input limits in the deployment guide). Report any input that causes excessive CPU/memory as a vulnerability.
  • No telemetry. The core makes no network calls and sends us nothing.

There aren't any published security advisories