The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides logic for processing external maritime data such as AIS (Automatic Identification System) messages and cargo manifests. This creates a potential surface for indirect prompt injection where malicious instructions could be embedded in external data fields to influence the agent's actions.
Ingestion points: Data enters the context through ais_data in the process_ais_message method (references/VESSEL_TRACKING_SYSTEM.md) and container/manifest lists in the calculate_stowage_plan and track_bill_of_lading methods (SKILL.md).
Boundary markers: The skill does not define specific delimiters or instructions to the agent to disregard natural language instructions that might be found within the maritime data fields.
Capability inventory: The skill is configured with Read, Write, and Edit tools, which provides the agent with the ability to modify the environment based on the processed information.
Sanitization: The reference code implements domain-specific validation (e.g., checking vessel status and speed) but does not include sanitization to prevent the LLM from interpreting data values as executable instructions.

maritime-expert