The Agent Skills Directory

[SAFE]: The skill offers standard frontend development guidelines for handling broken images and improving Core Web Vitals. The instructions align with industry best practices for error handling and performance optimization.
[INDIRECT_PROMPT_INJECTION]: The skill involves scanning external codebase files (HTML/JSX/CSS) to identify image paths. While this constitutes processing untrusted data, the skill's capabilities are limited to reporting missing assets and suggesting standard attribute-based fixes. There are no triggers for network operations, privilege escalation, or arbitrary command execution based on the ingested content.
[COMMAND_EXECUTION]: The provided Node.js example (scripts/check-images.js) uses standard file system modules (fs, path) and the glob utility to validate local file paths. This is a typical CI/CD utility for project maintenance and does not involve executing untrusted input or accessing sensitive system paths.

broken-images