The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a diagnostic script (scripts/equities_stocks_diagnostics.py) that ingests external data from a CSV file.
Ingestion points: Data is loaded from a user-provided CSV file via pandas.read_csv in scripts/equities_stocks_diagnostics.py.
Boundary markers: Not applicable as the data is processed numerically rather than interpolated into prompts.
Capability inventory: The script performs file reads and writes (via the --output flag) but lacks network access or arbitrary command execution capabilities.
Sanitization: The script uses pd.to_numeric(errors='coerce') to ensure input data is treated as numbers, effectively filtering out non-numeric injection attempts.
[COMMAND_EXECUTION]: The skill instructions advise running a local Python script for diagnostics. This is a standard and safe use of included scripts for data analysis.
[DATA_EXPOSURE]: The diagnostic script processes portfolio data but does not access sensitive system paths (e.g., .ssh, .aws) or perform network operations that could lead to exfiltration.

equities-stocks