The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The script scripts/credit_relative_value_diagnostics.py ingests user-provided data via the input.csv file.
Ingestion points: The pd.read_csv function in scripts/credit_relative_value_diagnostics.py reads external CSV content into a DataFrame.
Boundary markers: None present; data is processed as raw input.
Capability inventory: The script is limited to reading the input file, performing mathematical calculations using pandas, and writing the results to a local JSON file or stdout. It does not contain eval(), exec(), or network capabilities.
Sanitization: The script uses pd.to_numeric and pd.to_datetime with errors='coerce', which effectively filters out non-numeric and non-date content from the analysis.
[COMMAND_EXECUTION]: SKILL.md provides instructions to execute scripts/credit_relative_value_diagnostics.py. This execution is localized to the skill's own directory, processes user-supplied numerical data, and does not involve privilege escalation or shell injection risks.

credit-relative-value