huashu-material-search

Warn

Audited by Gen Agent Trust Hub on Feb 25, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides explicit instructions for the agent to execute shell commands using the grep utility. It details specific command-line arguments such as -i for case insensitivity, -n for line numbering, and -C for context display to be run on the host system.
  • [DATA_EXFILTRATION]: The skill is configured to access and read from a hardcoded absolute path on the user's local filesystem: /Users/alchain/Documents/写作/写作参考/个人素材库/全部即刻动态.csv. This involves processing and potentially exposing private content from the user's document directory to the AI model context.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 25, 2026, 09:38 AM