The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains no instructions designed to bypass safety filters or override agent behavior. The text is purely educational and focused on technical implementation details for web images.
[DATA_EXPOSURE]: No hardcoded credentials, API keys, or sensitive file paths were detected. The skill uses standard data URI patterns for image placeholders, which do not contain sensitive information.
[OBFUSCATION]: Several Base64 strings are present in references/modern-formats.md and references/lazy-loading.md. These were analyzed and confirmed to be legitimate binary data for tiny WebP/AVIF images used for feature detection and SVG placeholders. They do not decode to executable commands or hidden instructions.
[REMOTE_CODE_EXECUTION]: No patterns for remote code execution were found. The skill suggests using established developer tools like sharp (via npm) and Lighthouse (via npx), which are standard industry practices.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a code generation assistant. While it ingests user-provided strings for image attributes (src, alt), these are interpolated into standard HTML/React templates. The risk is negligible as these are standard developer inputs and the skill itself lacks dangerous capabilities like arbitrary network transmission or sensitive file reads.

responsive-images