The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local bash scripts to handle the deployment of generated HTML files to Vercel, AWS S3, and GCP/Firebase.
Evidence: SKILL.md references execution of deploy.sh scripts located in skills/vercel-deploy/, skills/aws-deploy/, and skills/gcp-deploy/.
[EXTERNAL_DOWNLOADS]: The skill references assets from trusted organizations and well-known services for typography and iconography.
Evidence: The skill includes CDN references to fonts.googleapis.com (Google Fonts), unpkg.com (Lucide icons), and cdn.jsdelivr.net (Phosphor and Tabler icons).
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it transforms untrusted user descriptions into production HTML content.
Ingestion points: User descriptions are parsed during the 'Gather Context' phase in SKILL.md.
Boundary markers: No delimiters or specific instructions are provided to the agent to treat user input as untrusted data during page generation.
Capability inventory: The skill has shell execution capabilities for site deployment across multiple cloud providers.
Sanitization: The instructions do not define any sanitization or validation logic for the user-supplied content before it is embedded in the index.html file.

landing-page-builder