The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions utilize the node command to execute a series of provided JavaScript scripts (e.g., init-clone.mjs, recon-site.mjs, mirror-site.mjs). These scripts manage the cloning lifecycle, including directory creation, browser-based reconnaissance, and file auditing.
[EXTERNAL_DOWNLOADS]: The skill performs numerous network operations to fetch remote content. Tools like gh api are used to search GitHub repositories, while curl, wget, and the Node.js fetch API are used to download website assets, source maps, and static files from user-specified URLs.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests and processes untrusted data from the web.
Ingestion points: Data enters the environment via target URLs processed by recon-site.mjs, route-crawl.mjs, and mirror-site.mjs.
Boundary markers: Absent. The skill extracts titles, headings, and metadata from target websites and incorporates them directly into local JSON and Markdown reports.
Capability inventory: The skill can execute local Node.js scripts, write files to the local file system (within ~/projects/website-clones/), and perform network downloads.
Sanitization: The scripts perform basic text cleaning (e.g., whitespace removal, slicing) for reporting purposes, but do not implement prompt-specific sanitization or escaping for the ingested content.

web-clone