The Agent Skills Directory

[SAFE]: The skill's primary functionality is to orchestrate a series of UI audits using local configuration files and internal skill dependencies. No suspicious command execution or network exfiltration patterns were found.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted design descriptions and screenshots, creating a potential surface for indirect prompt injection.
Ingestion points: The design_description input field in SKILL.md accepts external user data.
Boundary markers: The instructions do not specify explicit delimiters to isolate user-provided content from the agent's logic.
Capability inventory: This orchestration skill calls sub-skills but lacks direct high-risk capabilities like network exfiltration or sensitive file system modifications.
Sanitization: No specific input validation or sanitization routines are mentioned for the design description data.

refactor-ui