csv-data-summarizer
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains imperative instructions in SKILL.md, such as '⚠️ CRITICAL BEHAVIOR REQUIREMENT ⚠️' and 'DO NOT ASK THE USER WHAT THEY WANT', which are intended to override standard agent safety protocols and conversational patterns.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted CSV data and incorporates it into the model's response without sanitization or clear boundaries.\n
- Ingestion points: Data is ingested through pandas.read_csv in analyze.py using a user-provided file path.\n
- Boundary markers: There are no markers or instructions provided to the agent to treat the data as untrusted or to ignore embedded commands.\n
- Capability inventory: The script is capable of writing files to the local filesystem (e.g., plot images).\n
- Sanitization: No validation or escaping is performed on column names or data values before they are returned to the agent context.
Audit Metadata