Skip to content

Security: SkillBench-AI/skillmeter-plugins

Security

SECURITY.md

Security Policy

Supported version

This policy covers both SkillMeter clients in this repository: SkillMeter for Claude Code and SkillMeter for Codex. Security fixes are applied to the latest released version of each client. Update to it before reporting an issue that may already be resolved.

Report a vulnerability

Send security and privacy reports to privacy@skillbench.com with the subject [SkillMeter Security]. Do not open a public issue for a suspected vulnerability.

Include:

  • the affected client (Claude Code or Codex), its SkillMeter version, and the Claude Code or Codex version;
  • the operating system;
  • a minimal reproduction using synthetic data;
  • expected and observed behavior; and
  • the potential privacy or security impact.

Do not send GitHub access tokens, SkillMeter license JWTs, raw transcripts, customer source code, generated telemetry logs, or other production secrets. If sensitive evidence is necessary, first ask for a secure transfer method.

SkillBench will investigate reports with reasonable care and may ask for additional reproduction details. This repository does not currently operate a public bug-bounty program.

Security boundaries

  • Plugin hooks run with the user's local permissions.
  • Telemetry requires a valid license plus explicit organization and repository authorization.
  • Network requests use HTTPS and authenticated endpoints.
  • Secret redaction is defense in depth and is not a guarantee that arbitrary content is free of sensitive information.
  • Local telemetry queues can contain sanitized but still confidential content and should be protected like other developer data.

Disclosure

Please allow reasonable time for investigation and remediation before public disclosure. SkillBench may coordinate disclosure timing when users need an update before details are published.

There aren't any published security advisories