prototype-options
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads data from
.scratchpad/<slug>/interview-decisions.mdto build prototypes. Malicious instructions previously injected into this file could influence the agent during the prototype generation phase. - Ingestion points: Processes content from
.scratchpad/<slug>/interview-decisions.md. - Boundary markers: No explicit delimiters or instructions are provided to the subagents to ignore potentially malicious content within the decision file.
- Capability inventory: The skill allows writing files to the filesystem and executing small scripts or components.
- Sanitization: No evidence of sanitization or validation of the decision data before it is used to generate code.
- [DYNAMIC_EXECUTION]: The skill generates and executes code (scripts, components, or HTML) dynamically based on user requirements to create navigable prototypes.
- Evidence: The instructions explicitly suggest using "a small script or a single component" to demonstrate options, which involves creating and then running or rendering new code at runtime.
Audit Metadata