deepstream-dev

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading pre-trained models and tools from trusted sources.
  • Fetches models from the NVIDIA NGC container registry (nvcr.io).
  • Downloads utility scripts from official NVIDIA GitHub repositories (github.com/NVIDIA-AI-IOT).
  • These operations are documented for standard development workflows.
  • [COMMAND_EXECUTION]: The skill includes shell commands for environment setup and application deployment.
  • Instructions for installing the SDK components using local .whl files and apt-get for system dependencies (librdkafka-dev, libmosquitto1).
  • Commands for running official DeepStream Docker containers with GPU and X11 display support.
  • All commands are transparent and aligned with the skill's primary purpose of SDK development.
  • [DATA_EXFILTRATION]: The skill documents intended data transfer capabilities to external systems.
  • Provides code patterns for streaming inference metadata to Kafka brokers using the nvmsgbroker plugin.
  • Includes examples of streaming frames to REST endpoints using the Python requests library.
  • These are clearly defined features for video analytics workflows and do not involve unauthorized data access.
  • [PROMPT_INJECTION]: No malicious prompt injection patterns were found. The instructions use standard instructional language to guide the AI agent in generating correct code.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 04:34 PM
Security Audit — agent-trust-hub — deepstream-dev