data-analysis-skill
Warn
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions (SKILL.md) include commands to download and install the full skill implementation from a third-party GitHub repository:
https://github.com/dongzhang84/data-analysis-skill. This repository is not verified. - [REMOTE_CODE_EXECUTION]: The skill documentation explicitly states that its bundled scripts will perform automatic installation of Python packages (
pandas,openpyxl,tabulate,chardet) usingpipon the first run. This behavior involves executing commands that download and install external code without explicit user confirmation of the specific package versions or sources. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external datasets (CSV and Excel files) using a "multi-expert parallel analysis" approach. This creates a vulnerability surface where instructions embedded within the processed data files could influence the behavior of the subagents.
- Ingestion points: Reads Excel and CSV datasets provided by the user.
- Boundary markers: None mentioned in the instructions for delimiting data from subagent instructions.
- Capability inventory: Multi-expert parallel analysis, HTML report generation with interactive charts, and PowerPoint export capability.
- Sanitization: No evidence of input sanitization or validation of the contents of the Excel/CSV files before processing.
Audit Metadata