The Agent Skills Directory

[SAFE]: The skill implements a legitimate workflow for financial quantitative research and data quality monitoring.
[DATA_EXPOSURE]: The diagnostic script scripts/financial_data_science_diagnostics.py reads user-supplied CSV files and generates statistical summaries (mean, standard deviation, quantiles). It writes results to a local JSON file as specified by the user. There are no network calls, hardcoded credentials, or access to sensitive system paths.
[DYNAMIC_EXECUTION]: The Python implementation uses standard libraries (pandas, json, argparse) and avoids dangerous functions such as eval(), exec(), or runtime code generation.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data through the input.csv file. However, the processing is limited to numeric and timestamp columns, with explicit type conversion (pd.to_numeric, pd.to_datetime). This strict parsing significantly mitigates the risk of external data being interpreted as instructions by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to run its own bundled Python script. This is the intended primary purpose of the skill and does not involve unauthorized shell execution or privilege escalation.

financial-data-science