The Agent Skills Directory

[SAFE]: The skill implements its logic through several local scripts (e.g., ea-parse.ts, calibre-db.ts) that perform deterministic file and database operations. These operations are transparently defined and restricted to the local library environment specified by the user.
[COMMAND_EXECUTION]: The skill provides numerous Deno scripts intended to be executed with specific permissions (e.g., --allow-read, --allow-write). While some use the -A (all permissions) shebang, this is common for local developer tools and is used here solely to manage the local knowledge base and process ebook files as described in the documentation.
[PROMPT_INJECTION]: The extraction workflow involves processing external ebook content through LLM prompts. While the prompts use structured sections to delineate source text, they lack explicit boundary markers or warnings against instruction injection within the books. This presents a potential surface for indirect prompt injection, which is documented as a low-severity inherent risk for skills performing text analysis.
[EXTERNAL_DOWNLOADS]: Several scripts import standard utility modules from well-known and trusted registries such as deno.land and jsr.io. No other remote code patterns or suspicious external network activity was found.

ebook-analysis