The Agent Skills Directory

[PROMPT_INJECTION]: The skill describes a workflow for scraping existing websites (specifically those built on Wix) using browser automation tools. This creates an attack surface for indirect prompt injection, where malicious instructions hidden in a third-party website's content could influence the agent's behavior.
Ingestion points: The section 'Scraping & Rebuilding Existing Sites' directs the agent to use browser tools like get_page_text and javascript_tool on external, untrusted URLs.
Boundary markers: The instructions do not define clear boundaries or 'ignore' directives to prevent the agent from following instructions embedded within the scraped content.
Capability inventory: The skill possesses the ability to execute network requests, run JavaScript, and write files to the local environment.
Sanitization: No explicit sanitization or filtering of the scraped content is mandated before the data is processed by the agent.
[COMMAND_EXECUTION]: The skill provides instructions for deploying website files to the Netlify API using shell commands like zip and curl.
Evidence: The 'Deployment (Netlify via API)' section provides a script to bundle files and transmit them to api.netlify.com.
[DATA_EXFILTRATION]: The skill establishes a pattern for using API tokens to authenticate with remote services. While it uses placeholders, the workflow involves transmitting sensitive authentication material to a remote endpoint.
Evidence: The deployment scripts demonstrate the use of a NETLIFY_TOKEN variable to authorize network requests to external infrastructure.

website