scenario-walkable-room
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references the installation of additional components from the vendor's repository using
npx skills add scenario-labs/skills. It also utilizes well-known third-party libraries and tools includingthree.js,Spark,Rapier, andffmpegfor 3D rendering, physics, and audio processing. - [DYNAMIC_EXECUTION]: The instructions guide the agent to perform local data processing and spatial calibration using Python scripts. These scripts perform geometric calculations (RANSAC, least-squares fits) on downloaded Gaussian splat data to establish floor levels and scales.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided photos as the primary data source. While any external data ingestion presents a theoretical attack surface for vision-based instruction injection, the skill uses this data specifically for visual description and 3D reconstruction tasks within a defined pipeline.
Audit Metadata