The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests complex JSON data structures containing property information, owner psychology profiles, and valuation details. These free-text fields are processed by the agent to generate reports and strategy recommendations. If the input JSON files originate from an untrusted third party, they could contain embedded instructions designed to influence the agent's output.
Ingestion points: Input data is loaded via load_input_data() in negotiation_strategy_planner.py and processed in the NegotiationSettlementCalculator class in negotiation_settlement_calculator.py.
Boundary markers: The skill does not implement explicit boundary markers or 'ignore embedded instructions' warnings when processing natural language fields in SKILL.md or the output modules.
Capability inventory: The skill possesses the capability to perform file-write operations (generating reports and JSON results) and execute complex analysis based on the ingested data.
Sanitization: While modules/validators.py provides robust type and range validation, it does not include sanitization or filtering for potential prompt injection patterns within descriptive text fields.

negotiation-expert-infrastructure