suggest-consulting-packages

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes restricted Bash commands (mkdir, test, date) as specified in the allowed-tools frontmatter. These commands are constrained to creating directories, verifying file existence, and generating timestamps for file naming, which are standard operations for managing localized output files.
  • [DATA_EXPOSURE]: The procedure involves reading a local file named ground-truth.md to retrieve context regarding the user's 'freelance-brand'. This file access is limited to the defined working context of the agent and is necessary for the skill's primary function of generating personalized recommendations.
  • [EXTERNAL_DOWNLOADS]: The skill employs WebSearch and WebFetch to gather information about targeted domains or companies. This usage is transparent and driven by user-supplied arguments to provide relevant research, with no evidence of automatic remote script execution or data exfiltration to unauthorized endpoints.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 05:16 AM
Security Audit — agent-trust-hub — suggest-consulting-packages