Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/wa.pyscript utilizesos.execvto perform process replacement, ensuring it runs under a specific Python interpreter within its managed virtual environment. It also executes the localscripts/_bootstrap.shfile usingsubprocess.runto handle initial environment setup.- [EXTERNAL_DOWNLOADS]: Thescripts/_bootstrap.shscript installs thepywhatsandqrcodelibraries from the official Python Package Index (PyPI).- [CREDENTIALS_UNSAFE]: To maintain its connection as a companion device, the skill stores WhatsApp authentication sessions and Signal protocol databases in the user's~/.pywhatsdirectory. This is intended functionality for the skill's messaging capabilities.- [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection through thelistensubcommand which processes incoming WhatsApp messages. - Ingestion points: External messages received via
scripts/wa.py listeninscripts/wa.py. - Boundary markers: Not present in the JSON-formatted event stream.
- Capability inventory: Messaging commands in
scripts/wa.pyincludingsend-text,send-image, andgroup-send. - Sanitization: Not present.
Audit Metadata