The Agent Skills Directory

[PROMPT_INJECTION]: The file skill-report.json contains a self-referential security analysis that asserts a 'safe' verdict and pre-emptively explains away potential scanner findings as false positives. This technique is designed to influence or bias automated security reviews and is classified as a low-severity indirect prompt injection (Category 8e).
[SAFE]: The core content of the skill is high-quality documentation covering frontend architecture, accessibility, and security for major frameworks. It contains no executable scripts, shell commands, or network-active components.
[SAFE]: The skill recommends several reputable development and security tools, such as ESLint, Snyk, and Socket. These references are provided as optional improvements for the developer's environment and are presented as recommendations for human review and adoption.

frontend-guardrails