codex-storyboard-video-production

Warn

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill directs users to download and install a plugin from a personal GitHub repository (Yuuhann1999/codex-storyboard) that is not associated with a verified organization or the stated vendor.
  • [REMOTE_CODE_EXECUTION]: The standalone installation guide instructs users to clone an external repository and run 'npm start', which executes arbitrary third-party code on the local system.
  • [COMMAND_EXECUTION]: Installation via the 'codex plugin' marketplace involves downloading and executing packages from a third-party registry, which can lead to the introduction of unverified code.
  • [DATA_EXFILTRATION]: The 'complete_storyboard_generation_task' tool utilizes the 'file://' protocol to handle media assets. This creates a potential data exposure surface where the agent could be directed to access or reference sensitive local system files if the path parameters are manipulated.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the ingestion of external data files.
  • Ingestion points: The agent reads 'DESIGN.md' guidelines and 'project.json' scripts to inform asset generation tasks.
  • Boundary markers: There are no instructions or delimiters provided to ensure the agent ignores or sanitizes instructions embedded within these files.
  • Capability inventory: The skill has the ability to delete project files, update local data, and invoke external media generation services.
  • Sanitization: The skill lacks any mentioned validation or filtering for the content of processed design documents or scripts.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 8, 2026, 12:41 AM
Security Audit — agent-trust-hub — codex-storyboard-video-production