calibrate-metropolis-camera

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prim paths and configuration metadata which serves as an ingestion point for potentially untrusted data that could influence the agent's behavior.
  • Ingestion points: The parameters scene_root, camera_parent_prim_path, and place_info in scripts/calibrate_camera.py are supplied by the user at runtime.
  • Boundary markers: Absent; the skill relies on the underlying Isaac Sim Kit APIs to handle these strings.
  • Capability inventory: The skill can write calibration JSON and rendered images to the local filesystem and enable extensions via the omni.kit.app extension manager.
  • Sanitization: The script validates that the output_dir does not contain a URI scheme (e.g., ://) to prevent remote data exfiltration and verifies the format of the place_info configuration string.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 02:58 PM
Security Audit — agent-trust-hub — calibrate-metropolis-camera