The Agent Skills Directory

External Content Processing (Indirect Ingestion): The skill is designed to capture and replicate the structure and assets of user-specified URLs. While this is a core feature, processing untrusted web content is a standard consideration for browser-enabled tools.
Ingestion points: External data enters the agent context via the <target-url> during the browser capture phase.
Boundary markers: The skill includes a mandatory warning step (Step 1) requiring the user to acknowledge terms of service.
Capability inventory: The workflow allows for file writes, shell command execution via npm, and network previewing through sites-preview.
Sanitization: Verification steps (Step 10 and 12) require manual inspection of the rendered page and interaction testing to ensure integrity.
Local Development Command Execution: The skill uses command-line tools to initialize and run the generated prototypes. This is a standard part of a developer workflow.
Evidence: It utilizes shell commands such as npm install, npm run dev, and sites-preview to build and serve the project.
Context: These operations are typical for modern web development environments. The process includes verification checks to pass design QA before completion.
Grounding Material Access: The agent is instructed to use local context such as design systems and tokens to improve cloning accuracy.
Evidence: Instructions reference $user-context, codebase paths, and tokens as grounding material.
Context: This behavior is intended to align the generated output with the user's existing codebase and professional design standards.

url-to-code