The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a startup instruction in SKILL.md that directs the AI agent to execute a directory listing command (ls) to check for the presence of recommended companion skills in specific local configuration directories.
[EXTERNAL_DOWNLOADS]: The documentation provides installation instructions using the npx package runner to download skills from the AbsolutelySkilled repository, which serves as the distribution point for this suite of tools.
[SAFE]: The skill is designed to ingest and process external datasets, representing a potential surface for indirect prompt injection attacks. This is an expected and necessary characteristic for data analysis functionality.
Ingestion points: Data is loaded from local CSV files using standard pandas methods (pd.read_csv).
Boundary markers: No specific delimiters or safety instructions are defined for processing input data files.
Capability inventory: The agent context allows for the execution of Python code and shell commands.
Sanitization: While the skill provides templates for data cleaning (handling missing values, outliers, duplicates), it does not include specific sanitization against embedded malicious instructions in data fields.

data-science