The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions include a mandatory workflow step to execute a Python helper script (run_stat_analysis.py) via the command line. This script processes user-supplied CSV files, which can lead to command injection if filenames or paths are not properly sanitized.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting and processing untrusted external data (CSV, JSON, XLSX) to generate narrative scientific insights and interpret research findings. It lacks explicit boundary markers or instructions to ignore embedded commands within the processed data.
Ingestion points: Reads project data files (.csv, .json, .xlsx) in SKILL.md workflow steps 1 and 2.
Boundary markers: None identified; there are no instructions to the agent to disregard instructions potentially hidden in the dataset content.
Capability inventory: Executes shell commands (python run_stat_analysis.py), generates and runs Python/R scripts, and utilizes the computer-use tool to automate GUI applications.
Sanitization: Not present; the statistical helper script reads raw CSV data and calculates metrics without validation for adversarial content.
[DYNAMIC_EXECUTION]: The workflow involves generating and executing custom Python/R scripts (.py or .ipynb) based on data characteristics. While standard for a data science agent, it constitutes runtime script generation and execution.

data-scientist-analyst