csv-workbench

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFE
Full Analysis
  • Command Execution: The skill uses shell-templated Python scripts to analyze data. This allows the agent to perform complex calculations efficiently within the local environment.
  • Indirect Prompt Injection Surface: The skill is designed to analyze external data which could potentially contain instructions intended to influence the agent. 1. Ingestion points: CSV files located in /mnt/data/ (referenced in SKILL.md and playbook.md). 2. Boundary markers: The skill relies on standard CSV parsing but does not explicitly define markers to separate data from instructions. 3. Capability inventory: Includes shell command execution and Python script execution. 4. Sanitization: Standard CSV parsing via the Python stdlib is utilized. This surface is a normal part of data analysis tools and is managed through standard agent safety protocols.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 10:25 AM