The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes US customs shipment records (bills of lading) which contain free-text product descriptions provided by external filers. This data is ingested into the agent's context, creating a potential surface for indirect prompt injection if a filer includes malicious instructions.
Ingestion points: Output from the trade command-line tool executed via the bash tool (SKILL.md).
Boundary markers: The skill does not define specific delimiters (e.g., XML tags or Markdown blocks) or instructions for the agent to ignore embedded commands within the shipment data.
Capability inventory: The agent has access to the bash tool for command execution and the create_workbook tool for automated research (SKILL.md).
Sanitization: There are no instructions for sanitizing, escaping, or validating the text returned from the trade search before it is processed by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to use a bash tool to interact with the trade utility and check for available programs via programs list. While this is intended functionality for a research agent, shell access represents a broad capability that requires careful monitoring.

us-importers