The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the 30book-mcp package using the uvx tool. This is a standard mechanism for extending agent capabilities via the Model Context Protocol (MCP).
[INDIRECT_PROMPT_INJECTION]: The skill tools ingest data from external sources, specifically book metadata and reader reviews (e.g., sb_reviews). This constitutes an attack surface for indirect prompt injection if the external content were to contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data is retrieved via tools like sb_search, sb_book, and sb_reviews from site APIs.
Boundary markers: The skill does not explicitly define delimiters for the retrieved content within this document.
Capability inventory: The tools are described as read-only and intended for information retrieval; no file-writing or system-command capabilities are exposed to the processed data.
Sanitization: No explicit sanitization or filtering logic is mentioned in the instruction file.

30book