startup-validation-test-designer
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The instructions are strictly limited to business validation workflows and do not contain any patterns attempting to override agent safety filters or system instructions.
- [DATA_EXPOSURE]: The skill includes an 'Ethics, privacy, and sensitive-domain guardrails' section that explicitly directs the agent to collect the minimum data needed and avoid sensitive categories such as financial, health, or legal data.
- [DYNAMIC_EXECUTION]: While the skill allows for the creation of an HTML artifact, it includes mandatory security instructions to use self-contained files without external dependencies and to 'HTML-escape user-provided text and do not execute or embed user-provided HTML or script' to prevent injection attacks.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a clear ingestion surface for user-provided startup problems and assumptions. It mitigates injection risks by instructing the agent to process these inputs into a structured output format and provides explicit sanitization instructions for the artifact generation process.
Audit Metadata