The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to embed a script from https://cdn.inlay.dev/embed.js into their website's HTML or layout files. This leads to the execution of remote code on the user's infrastructure from a source that is not in the trusted vendors list.
[COMMAND_EXECUTION]: The workflow includes multiple terminal commands using curl to interact with mcp.inlay.dev and inlay.dev/api/audit. These are provided for the user to manually verify and test the MCP tools and their site's AI readiness.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of an MCP server that indexes and retrieves website content for AI agents, which creates a vulnerability where malicious content on the website could influence agent behavior.
Ingestion points: Website content retrieved via the search_pages and get_page_content tools following indexing by the Inlay embed script.
Boundary markers: The skill does not provide specific instructions for using delimiters or boundary markers to separate site content from agent instructions.
Capability inventory: The tools allow for metadata retrieval, content searching, and full page content extraction.
Sanitization: No sanitization or filtering logic is described in the skill for handling the content retrieved from the external website.

setup-mcp-server