dreamina-video-evaluator
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill contains a data ingestion surface for external measurement data. 1) Ingestion points: Reads clip gates and design intent (SKILL.md). 2) Boundary markers: Present in the 'Rules' and 'Workflow' sections of SKILL.md and references/workflow-contract.md. 3) Capability inventory: Restricted to reading inputs and producing a specific JSON output (SKILL.md). 4) Sanitization: Mandatory requirement to label missing evidence as 'NOT_VERIFIED' (references/workflow-contract.md).
- [NO_CODE]: The skill consists entirely of markdown instructions and YAML configuration, with no executable scripts, binaries, or automated shell commands.
- [SAFE]: The skill implements proactive security protocols, including explicit prohibitions against logging secrets or sensitive personal data and a requirement for human authorization for any state-changing operations.
Audit Metadata