Skip to content

Security: krasnoperov/patchrelay

Security

SECURITY.md

Security Policy

Reporting vulnerabilities

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

Report vulnerabilities privately to the maintainers with:

  • a description of the issue
  • affected versions or commits
  • reproduction steps or proof of concept
  • any mitigations you have already identified

We will acknowledge the report, investigate it, and coordinate a fix and disclosure plan.

Deployment posture

PatchRelay is designed for infrastructure you control. Recommended production posture:

  • bind to loopback unless you have a strong reason not to
  • expose only /, /health, /ready, and POST /webhooks/linear
  • leave operator_api.enabled disabled unless you explicitly need the HTTP operator endpoints
  • require PATCHRELAY_OPERATOR_TOKEN if you enable the operator API on a non-loopback bind
  • treat workflow files and Codex runtime access as privileged automation policy
  • configure projects[].trusted_actors so only trusted Linear owners or trusted domains can trigger automation

Configuration model

PatchRelay separates trust into three layers:

  • Network trust: the public server surface should only accept signed Linear webhooks.
  • Operator trust: local setup and inspection routes are for the operator, not the public internet.
  • Linear actor trust: a valid webhook is still ignored if its actor is not trusted for the routed project.

trusted_actors is evaluated per project. When configured, unmatched issue and comment events are ignored before they can change desired stage state or inject comment context into an active turn.

There aren’t any published security advisories