nan-ebook-download

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt instructs the agent to extract a session-bound get.php URL containing a key (https://libgen.li/get.php?md5=MD5&key=KEY) and to use it immediately in terminal download commands, which requires the LLM/agent to capture and emit that token verbatim (exfiltration risk).

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Several entries are shadow-library / mirror / CDN download endpoints (libgen.*, annas-archive, oceanofpdf, booksdl CDN, Mobilism) that serve untrusted, session-keyed files and user uploads — a common vector for malicious or stubbed downloads — even though the list also contains legitimate sources (Gutenberg, Archive.org, Open Library, Google Books).

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged a literal, random-looking session key that grants access to a libgen get.php download URL. The string "TD4SIA4HAG3C4381" is high-entropy (uppercase alphanumeric) and appears inline as the GET key parameter for a download URL — a value that provides access to a resource when combined with the md5. Even if session-bound / likely expired, it is a real token embedded in the docs and meets the "literal credential" rule.

I did NOT flag MD5 hashes (they are identifiers, not secret credentials), local proxy addresses (127.0.0.1:7897), or other example strings and config placeholders (these are low-entropy or clearly documentation examples).