The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill utilizes a Python calculator that ingests untrusted external data, creating a potential vector for indirect prompt injection.
Ingestion points: The load_expropriation_from_json function in expropriation_calculator.py reads user-provided JSON files from the local filesystem via command-line arguments.
Boundary markers: The skill does not implement explicit boundary markers or instructions to the model to ignore embedded commands within the processed data.
Capability inventory: The skill includes file-writing capabilities through the save_results_to_json function, which writes analysis results to a path specified by the user.
Sanitization: While the script performs data type validation and range checking for numeric values (e.g., market value and interest rates), it does not sanitize string inputs (such as property addresses or notes) which could be used to influence downstream agent behavior when the results are reviewed.

expropriation-compensation-entitlement-analysis