analyzing-text-with-nlp

Warn

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill features deceptive metadata and documentation. Although described as an NLP tool for tasks such as sentiment analysis and keyword extraction, the provided script scripts/analyze_text.py actually performs recursive directory traversal and file metadata collection using rglob('*'). This discrepancy could lead an agent to perform unintended reconnaissance of the host file system.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it is designed to ingest and analyze potentially untrusted external content.
  • Ingestion points: Text, code, and documents provided by users or external sources for NLP analysis as specified in SKILL.md.
  • Boundary markers: There are no delimiters or 'ignore' instructions present to help the agent distinguish between data to be analyzed and instructions embedded within that data.
  • Capability inventory: The agent is granted access to powerful tools including Read, Bash(cmd:*), Grep, and Glob.
  • Sanitization: The skill lacks any mechanism for input validation or the filtering of embedded instructions in the text being processed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 7, 2026, 05:30 AM