The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from user-provided URLs, which serves as a potential attack vector for indirect instructions. * Ingestion points: Website content retrieved via browser_navigate and browser_snapshot. * Boundary markers: None identified; content is extracted verbatim for specification generation. * Capability inventory: File system writing, shell command execution (npm/npx), and sub-agent dispatch with worktree isolation. * Sanitization: None identified for the extracted text or assets.
[COMMAND_EXECUTION]: The skill automates several shell commands to initialize and verify the project. Evidence includes the use of npx create-next-app@latest, npm run build, npx tsc --noEmit, and the execution of a custom Node.js script at scripts/download-assets.mjs.
[EXTERNAL_DOWNLOADS]: The skill is designed to programmatically fetch assets from remote target websites, including images, videos, and SVG elements.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses a local directory path outside its immediate scope, specifically referencing a template repository at ~/Projects/ai-website-cloner-template/.
[DYNAMIC_EXECUTION]: The skill implements a multi-agent orchestration pattern where it dynamically generates component specifications at runtime based on external data and passes them to sub-agents for code generation.

clone-website