The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data for validation, creating an attack surface for indirect prompt injection. * Ingestion points: Data enters through the validate_records function in scripts/validators.py, which accepts lists of dictionaries from external sources. * Boundary markers: Pydantic schemas provide structural boundaries, but natural language instructions within validated strings are not explicitly filtered. * Capability inventory: The skill allows access to Write, Edit, and Bash tools, which could be abused if validated data is passed to these tools without further sanitization. * Sanitization: Data is validated against types and regex (e.g., email format), but no content-level sanitization is present to prevent instruction injection.

pydantic-validation