visual-prompt-reverse-engineering
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill consists exclusively of Markdown files providing structured instructions for visual analysis and prompt synthesis. There are no scripts, binaries, or configuration files that execute code or manage dependencies.
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes user-provided images, creating a data ingestion surface. The evidence chain shows: (1) Ingestion points: reference images provided by the user; (2) Boundary markers: strict pipeline stages including Visual Fact Inventory, Master Visual Specification, and multiple audits; (3) Capability inventory: text generation only, with no tool access or system-level capabilities; (4) Sanitization: explicit evidence discipline requiring strictly observable facts and calibrated uncertainty.
- [SAFE]: No malicious patterns such as prompt injection, data exfiltration, obfuscation, or persistence mechanisms were detected.
Audit Metadata