ltx-2-video

Pass

Audited by Gen Agent Trust Hub on Jul 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill interacts with official LTX-2 resources including the hosted API (api.ltx.video) and official repositories on GitHub and Hugging Face (Lightricks/LTX-2). These external references are legitimate vendor endpoints and are handled with high-security standards in the provided implementation.
  • [COMMAND_EXECUTION]: The embedded Python script (ltx_hosted.py) executes ffprobe and ffmpeg to validate the container and stream properties of generated videos. These operations use subprocess.Popen with shell=False, enforce resource-limiting timeouts, and implement strict output buffering to prevent tool-based exploitation or resource exhaustion.
  • [PROMPT_INJECTION]: The inclusion of EVAL.md constitutes an indirect prompt injection surface. This file contains internal evaluation rubrics and test scenarios intended for the assessor. Exposing these administrative instructions to the AI agent provides it with the scoring logic and test cases, which can influence or bias the agent's behavior during task execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 12, 2026, 02:28 PM
Security Audit — agent-trust-hub — ltx-2-video