confidentiality-general

Pass

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE] (SAFE): The skill consists entirely of Markdown instructions and playbooks. No Python, JavaScript, or shell scripts are present.
  • [DATA_EXFILTRATION] (SAFE): No network-capable tools (curl, wget, fetch) or sensitive file access patterns were detected.
  • [COMMAND_EXECUTION] (SAFE): No subprocess calls or shell commands are included in the instructions.
  • [PROMPT_INJECTION] (SAFE): The instructions follow standard role-definition patterns for legal assistance and do not attempt to bypass safety filters or extract system prompts.
  • [INDIRECT_PROMPT_INJECTION] (SAFE): While the skill ingests user-provided contracts, it possesses no exploitable capabilities (such as network or file system access) that could be triggered by malicious content in the processed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 21, 2026, 11:45 AM
Security Audit — agent-trust-hub — confidentiality-general