The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external CSV files which presents an indirect prompt injection surface.
Ingestion points: User-provided Wise CSV statement files described in SKILL.md under the 'Workflow' section.
Boundary markers: The instructions do not define delimiters or specific safety instructions to the agent for handling text within the CSV fields (such as 'Description') that might be interpreted as instructions.
Capability inventory: Uses the categorize, transaction_search, and export_transactions tools mentioned in the SKILL.md 'Wilson Tools Used' section.
Sanitization: No sanitization or validation of the content of CSV fields is described beyond basic CSV header validation.

wise-import