prompt-engineering-multimodal
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides educational content and prompt templates without any associated scripts, binaries, or network-level instructions.\n- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on how to ingest and extract data from external multimodal files such as screenshots and audio recordings. While this introduces a potential attack surface for instructions embedded in external media, the skill addresses this by recommending structured JSON outputs, explicit task and constraint boundaries, and mandatory verification steps like the OCR-trust rule. It specifies ingestion points in prompt templates and identifies the use of file manipulation tools for reading artifacts, while providing guidelines for validation of extracted data.
Audit Metadata