Skills
Navigate
skills/isaac-sim/isaacsim/action-and-event-data-generation/Gen Agent Trust Hub

action-and-event-data-generation

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill coordinates workflows that ingest YAML configuration files used by LLM-backed components.
  • Ingestion points: Configuration files (e.g., my_scene.yaml) passed to actor_sdg.py or the AEDG application.
  • Boundary markers: The instructions do not specify explicit delimiters or 'ignore embedded instructions' warnings for the natural language processing components.
  • Capability inventory: The skill enables shell script execution (./isaac-sim.action_and_event_data_generation.sh) and Python-driven simulation runs.
  • Sanitization: The skill correctly utilizes yaml.safe_load to prevent arbitrary code execution during the parsing of configuration files, but it does not account for natural language instruction injection within the data content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 02:57 PM
Security Audit — agent-trust-hub — action-and-event-data-generation