The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill possesses a significant attack surface for indirect prompt injection. 1. Ingestion points: The load_data and load_csv methods in SKILL.md allow the ingestion of external data. 2. Boundary markers: There are no indicators of delimiters or instructions to ignore embedded commands within the ingested data. 3. Capability inventory: The skill has the capability to write files to the disk via generate_report (PDF/HTML) and various plot_* methods (PNG). 4. Sanitization: There is no evidence of sanitization or filtering of the input data before processing or inclusion in reports.
Command Execution (MEDIUM): The skill includes a CLI interface (statistical_analyzer.py) that executes commands based on user-provided arguments, including file paths and test parameters.
Data Exposure (LOW): The skill is designed to read local files (e.g., data.csv), which is necessary for its function but represents a potential vector for accessing sensitive local data if misdirected.
External Downloads (INFO): The skill relies on standard, trusted scientific Python libraries (pandas, scipy, statsmodels, etc.) listed in requirements.txt.

statistical-analyzer