The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions and templates are purely descriptive of design systems. No attempts to override agent behavior, bypass safety constraints, or extract system prompts were detected.
[DATA_EXFILTRATION]: No unauthorized network operations or access to sensitive local file paths (such as SSH keys or credentials) were found. The skill uses standard Google Fonts links (fonts.googleapis.com) for font substitution, which is a well-known and safe service.
[REMOTE_CODE_EXECUTION]: The skill does not perform any external package installations (npm/pip) or execute remote scripts. It instructs the agent to use standard tools like write_file to generate static HTML content.
[OBFUSCATION]: All templates are written in plain Markdown and standard CSS. No Base64-encoded payloads, hidden Unicode characters, or homoglyph attacks were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill involves processing a large number of design templates, these files are part of the skill package itself and do not ingest untrusted runtime data. The instructions provided to the agent for using these templates are consistent with the skill's stated purpose.

popular-web-designs