The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external CSV exports, which constitutes an indirect prompt injection attack surface if the agent is asked to analyze or debug malicious data files.
Ingestion points: External CSV files are read from the exports/ directory as specified in SKILL.md.
Boundary markers: The instructions do not include specific delimiters or guidelines for the agent to ignore or isolate potentially malicious instructions embedded within the CSV content.
Capability inventory: The skill has access to a broad set of capabilities including Read, Write, Edit, Bash, and Grep across all scripts.
Sanitization: The server implementation utilizes the csv-parse library for handling data but lacks explicit validation or sanitization of field contents before they are potentially processed by the agent.

fondo-local-dev-loop