csv-data-summarizer
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The SKILL.md file contains a section titled 'CRITICAL BEHAVIOR REQUIREMENT' that uses forceful, imperative language (e.g., 'DO NOT ASK', 'JUST DO IT', 'FORBIDDEN BEHAVIORS') to override the agent's default interactive and safety behaviors, compelling it to act proactively without user clarification.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted CSV data and incorporates results into its output without using boundary markers or 'ignore embedded instructions' directives, which could allow malicious data content to influence agent behavior.\n
- Ingestion points: analyze.py reads CSV files directly using pandas.\n
- Boundary markers: None present in SKILL.md.\n
- Capability inventory: Includes file system reads (CSV) and writes (PNG plots).\n
- Sanitization: No sanitization of CSV cell content is performed before processing.\n- [COMMAND_EXECUTION]: The skill executes a Python script (analyze.py) that interacts with the local file system by reading user-supplied CSV files and writing multiple image files to the workspace.
Audit Metadata