The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes magick for image processing and standard filesystem tools like cp and mkdir. It follows security best practices by using shell quoting ("$1") for user-provided input paths, which helps prevent command injection.
[PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection (Category 8).
Ingestion points: The skill reads application metadata from project files including package.json, site.webmanifest, and config/application.rb (Step 3).
Boundary markers: The instructions do not specify any delimiters or boundary markers to differentiate untrusted project data from the skill's own logic.
Capability inventory: The skill has broad capabilities to modify project source code (e.g., Rails application.html.erb and Next.js layout.tsx) and execute shell commands (Step 7).
Sanitization: The skill lacks requirements for sanitizing or escaping the extracted application name before it is interpolated into TypeScript metadata or HTML tags. A maliciously crafted project name could potentially result in code injection or malformed layout files during the update process.

favicon