The Agent Skills Directory

[PROMPT_INJECTION]: The skill's operational model introduces a surface for indirect prompt injection.
Ingestion points: The skill is auto-invoked whenever the agent edits .css files, which are untrusted external data sources.
Boundary markers: No specific instructions or delimiters are provided to help the agent distinguish between CSS code and potentially malicious instructions embedded in comments.
Capability inventory: The agent is authorized to use Read, Write, and Edit tools, providing a mechanism for file system manipulation if an injection is successful.
Sanitization: No methods for sanitizing or validating the input CSS content are defined in the skill.

css-author