dreamina-design-use
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-generated prompts to route tasks to specific image and video generation tools.
- Ingestion points: User instructions for text-to-image, text-to-video, and multimodal video generation as defined in the routing rules of SKILL.md.
- Boundary markers: The architecture requires an explicit ApprovalReceipt bound to a unique request fingerprint before any task submission, preventing autonomous execution of potentially malicious embedded instructions.
- Capability inventory: Includes CLI execution through DreaminaAdapter.run, state querying via OperationLedger, and media file downloads through ArtifactService.
- Sanitization: Input is validated by ImageService and VideoService to reject unsupported tokens, and every transaction is gated by the ApprovalGuard to ensure user confirmation.
- [DYNAMIC_EXECUTION]: The skill utilizes localized scripts for core functionality, such as scripts.dreamina_adapter for CLI bridging and scripts.artifact_service for media processing. It also includes an integrity check script, scripts/verify_skill_snapshot.py, which verifies that the local execution environment remains byte-identical to the verified source repository full-aigc-skills/dreamina-skills.
Audit Metadata