The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external CSV data through a diagnostic script (scripts/reference_data_corporate_actions_diagnostics.py) and instructions. If the input CSV contains malicious instructions in string fields (like source_name), these could influence the agent's behavior when it reviews the resulting diagnostic JSON or implementation memo.
Ingestion points: The skill reads external data from an input.csv file using pandas.read_csv in scripts/reference_data_corporate_actions_diagnostics.py.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the prompt instructions for the agent when handling the resulting artifacts.
Capability inventory: The agent has the capability to execute the diagnostic script via python, write JSON artifacts, and create markdown documentation.
Sanitization: The diagnostic script uses pd.to_numeric with errors="coerce" for metric columns, providing some type-level sanitization, but source_name and other potential string fields are not sanitized against prompt injection patterns.

reference-data-corporate-actions