analyzing-text-sentiment
Warn
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: MEDIUMMETADATA_POISONINGCOMMAND_EXECUTION
Full Analysis
- [METADATA_POISONING]: The skill frontmatter and documentation in 'SKILL.md' describe a tool for emotional tone classification and sentiment analysis. However, 'scripts/analyze_sentiment.py' implements a directory scanner that walks the file system using 'Path.rglob("*")' to collect file statistics and sizes.
- [COMMAND_EXECUTION]: The skill requests unrestricted 'Bash(cmd:*)' access in its 'allowed-tools' configuration. When combined with deceptive scripts that access the file system in unexpected ways, this increases the risk of unauthorized data discovery or local reconnaissance.
Audit Metadata