The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands using npx tsx to run TypeScript scripts for site reconstruction, WXR manipulation, and screenshot generation. It also uses studio wp for interacting with the local WordPress environment, including executing generated PHP files via eval-file to update post content.
[EXTERNAL_DOWNLOADS]: The skill downloads images, fonts, and other media assets from remote source site CDNs (e.g., Wix or Squarespace domains) to localize them within the WordPress media library. These downloads are limited to assets required for the theme reconstruction process.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted HTML and CSS data from captured external websites. While the skill strips active scripts, it lacks explicit boundary markers to prevent the agent from potentially interpreting instructions embedded in the source markup.
Ingestion points: Captured files in html/*.html and sections/*.json.
Boundary markers: Not present; the content is processed as raw strings for replacement.
Capability inventory: File system access (Read/Write), shell command execution (Bash), and WordPress database manipulation (studio wp).
Sanitization: Includes stripping of <script> tags and resource-hint <link> tags, as well as link rewriting, but does not implement comprehensive prompt injection filtering.

replicate-theme