If you discover a security vulnerability in this project, please report it through GitHub's private vulnerability reporting:
- Go to the Security tab of this repository
- Click Report a vulnerability
- Provide a description of the issue and steps to reproduce
We will acknowledge receipt within 3 business days and provide an initial assessment within 10 business days.
Do not open a public issue for security vulnerabilities.
This repository distributes AI coding plugins (skills and agents) as Markdown and JSON files. Security concerns include:
- Skills that instruct the AI to disable permissions or skip verification
- Bundled scripts with injection risks or unsafe command patterns
- Exposure of secrets, tokens, or credentials in skill definitions
- Reference files containing sensitive organizational data (PII, infrastructure details)
- Public Red Hat URLs (
ux.redhat.com,brand.redhat.com,patternfly.org) - Plugin manifest metadata (author names, descriptions)
Note: Internal URLs (.corp., pages.redhat.com, etc.) are blocked by CI — not because they're a vulnerability, but because they indicate content that belongs in the internal repo instead of this public one.
See GOVERNANCE.md for the review layers that every contribution passes through before it can affect a user's system.
Every PR is scanned by multiple automated tools at two enforcement levels:
Blocking (fails the PR):
- Internal URL & secret scan — grep-based detection of internal Red Hat URLs and token prefixes
Advisory (findings reported, does not block merge):
- CodeRabbit — AI-powered review for hardcoded secrets and unsafe code patterns (PR comments)
- Skillsaw — content linter that catches embedded secrets in skill definitions (annotations)
- AI Guardian — security scanner for credential exfiltration, prompt injection, and code vulnerabilities (SARIF)
Only the latest version on the main branch is supported.