The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection through its discovery workflow.
Ingestion points: The WebFetch tool is utilized in Phase 2 to retrieve content from external URLs ('Existing web presence') provided by the user during the discovery phase.
Boundary markers: Absent. The skill instructions do not specify the use of delimiters or provide instructions for the agent to ignore potentially malicious embedded commands within the fetched content.
Capability inventory: The skill has extensive permissions including Bash (shell execution), Write/Edit (file system modification), and Glob/Grep.
Sanitization: Absent. There are no instructions to sanitize, filter, or validate the content retrieved from external URLs before the agent processes and analyzes it.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of resources from well-known technology and cloud services.
Development dependencies such as tailwindcss, solid-js, and better-auth are installed from the official NPM registry.
Typography assets are loaded from Google Fonts (fonts.googleapis.com and fonts.gstatic.com).
Photographic assets are retrieved from Unsplash (images.unsplash.com).
An analytics script is incorporated from Datafast (https://datafa.st/js/script.js).
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform standard project initialization and management tasks, such as installing Node.js dependencies and configuring the Astro framework stack. This behavior is consistent with the primary purpose of a website-building agent.

website-design