The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from an external website (barracuda.com), which is a common vector for indirect prompt injection.
Ingestion points: Tools such as barracuda_com__read_page and barracuda_com__get_root (SKILL.md) fetch titles, text, and links from external pages.
Boundary markers: The instructions do not specify any delimiters or safety markers to help the agent distinguish between tool output and potential malicious instructions embedded in the web content.
Capability inventory: The skill includes tools for reading pages and performing browsing actions via the Unbrowse platform (e.g., unbrowse.browse.act in SKILL.md).
Sanitization: No explicit sanitization or filtering logic is documented for the content returned from the external site.
[EXTERNAL_DOWNLOADS]: The skill references a remote MCP server and API endpoints managed by the developer.
Evidence: The skill points to https://unbrowse.ai/mcp/barracuda.com for tool execution and https://unbrowse.ai/api/v1/sites/barracuda.com/openapi.json for REST access (README.md, SKILL.md, server.json). These resources are consistent with the vendor's own domain.

barracuda