Skip to content

Security: psenger/ai-agent-skills

SECURITY.md

Security Policy

Supported Versions

Only the latest release on main receives security fixes. Older versions are not patched.

Version Supported
2.x (latest)
< 2.0

Reporting a Vulnerability

Do not open a public GitHub issue for security vulnerabilities.

Use GitHub's private vulnerability reporting to submit a report confidentially:

Report a vulnerability

Your report is visible only to the maintainer. Include as much detail as you can:

  • Which skill or file is affected
  • Steps to reproduce or demonstrate the issue
  • Potential impact

Response Timeline

Milestone Target
Acknowledgement Within 5 business days
Triage and severity assessment Within 10 business days
Fix or mitigation Dependent on severity and complexity

Disclosure Process

  1. You submit a private advisory — only the maintainer can see it.
  2. The maintainer acknowledges receipt and assesses severity.
  3. A fix is developed and tested privately.
  4. A patched release is published to main.
  5. The advisory is published (coordinated disclosure). You will be credited unless you prefer to remain anonymous.

Scope

In scope:

  • SKILL.md files shipped in this repository
  • Skill scripts and reference files under skills/
  • The .claude-plugin/marketplace.json manifest

Out of scope:

  • Skills installed from this repo into your own environment — those run under your own Claude Code configuration and are your responsibility to audit.
  • Third-party tools or services invoked by skills (Claude API, GitHub API, etc.).
  • Vulnerabilities in Claude Code itself — report those to Anthropic.

There aren't any published security advisories