The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external websites which can lead to indirect prompt injection attacks.
Ingestion points: Data is ingested via tools such as capture_component, analyze_bundle, capture_live_site, and analyze_page_structure from any user-provided URL in SKILL.md.
Boundary markers: The instructions do not specify any delimiters or warnings for the agent to ignore instructions embedded within the scraped website data.
Capability inventory: The skill has the capability to generate and write HTML/CSS/JS files, create project structures using awwwards-build-project, and execute shell commands like npm install and npm run build in SKILL.md.
Sanitization: There is no mention of sanitizing or escaping the content retrieved from the external sites before it is used in the build process or passed to other tools.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands during the project conversion phase.
Evidence: Phase 4 explicitly requires running npm install && npm run build and awwwards-build-project in SKILL.md.

recreate