csv-analysis
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Prompt Injection (LOW): The
SKILL.mdfile contains directive instructions designed to bypass the agent's standard conversational flow. Patterns like 'CRITICAL BEHAVIOR REQUIREMENT', 'DO NOT ASK THE USER WHAT THEY WANT', and 'NO questions, NO options, NO waiting for user input' are used to force immediate action and override default helpfulness guidelines.\n- Indirect Prompt Injection (LOW): The skill is vulnerable to indirect prompt injection via the processing of untrusted CSV data.\n - Ingestion points: The
summarize_csvfunction inanalyze.pyreads a user-provided file path directly into a pandas DataFrame.\n - Boundary markers: No delimiters or warnings are used to prevent the agent from being influenced by malicious text contained within the CSV data.\n
- Capability inventory: The skill executes file read operations via
pd.read_csvand file write operations viaplt.savefigwithinanalyze.py.\n - Sanitization: There is no evidence of validation or sanitization of the CSV content, which could allow malicious strings to influence the agent's summary output.
Audit Metadata