zlib-download

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: During the installation process, the setup.sh script downloads a pre-compiled binary (annas-mcp) from a GitHub repository (iosifache/annas-mcp) and prepares it for execution on the local system.
  • [EXTERNAL_DOWNLOADS]: Fetches versioned binary releases from GitHub to support Anna's Archive features. This download from a well-known service is documented neutrally as it is a required dependency for the skill's functionality.
  • [COMMAND_EXECUTION]: The book.py script executes the downloaded annas-mcp binary using subprocess.run to perform book searches and downloads. The setup.sh script also executes shell commands to install dependencies and set file permissions.
  • [CREDENTIALS_UNSAFE]: Stores sensitive user information, including Z-Library login credentials and Anna's Archive API keys, in a local environment file (~/.claude/book-tools/.env).
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted book metadata from external sources.
  • Ingestion points: Search results and metadata retrieved from Z-Library and Anna's Archive APIs in book.py and Zlibrary.py.
  • Boundary markers: None implemented.
  • Capability inventory: Includes file system writing, network communication, and subprocess command execution.
  • Sanitization: Employs regular expressions to sanitize filenames and prevent path traversal or the use of illegal characters.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 01:45 PM