csv-data-analysis
Pass
Audited by Gen Agent Trust Hub on Mar 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The report template (report_template.html) fetches the ECharts library from cdn.jsdelivr.net and Tailwind CSS from cdn.tailwindcss.com. Both are well-known and trusted services for providing frontend assets.
- [COMMAND_EXECUTION]: The skill uses the execute_skill_script_file tool to run its own Python script (csv_analyzer.py) to extract statistical data. This is an intended and standard functionality for this skill.
- [PROMPT_INJECTION]: The skill processes user-uploaded data files. This creates an attack surface for indirect prompt injection or XSS through malicious data content (e.g., in column headers) being rendered in the report, though no active exploitation logic was detected. Evidence: 1. Ingestion points: csv_analyzer.py reads user files. 2. Capability inventory: execute_skill_script_file and html_interpreter. 3. Boundary markers: The skill instructs the agent to separate markers from insights. 4. Sanitization: The script casts most values to numeric types, but headers are treated as strings.
Audit Metadata