The Agent Skills Directory

[COMMAND_EXECUTION]: The operating rules include a fallback instruction for the user to execute npx skills add wondelai/skills/<slug> --global. This command leverages npx to download and run Node.js code from a remote source. This is categorized as a vendor-specific resource recommendation originating from the skill author's ecosystem.
[EXTERNAL_DOWNLOADS]: The use of npx for adding skills implies the download of external packages from a remote registry (npm).
[PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, such as visitor reviews, support tickets, exit surveys, and live website assets. This creates an attack surface for indirect prompt injection.
Ingestion points: Target website URLs, pasted live copy, CSS assets, and raw visitor feedback (surveys, reviews, support tickets).
Boundary markers: The instructions do not define explicit delimiters or security markers to isolate untrusted visitor content from the agent's core instructions.
Capability inventory: The skill possesses the ability to write files to the docs/ directory, suggest shell commands to the user, and orchestrate other internal skills.
Sanitization: There is no evidence of sanitization, filtering, or validation logic applied to the external content before it is processed or written to project documentation.

improve-website