sora
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill functions by directing the agent to execute a bundled Python CLI utility (
scripts/sora.py) via shell commands. This is the intended and documented design for interacting with the video generation API. - [EXTERNAL_DOWNLOADS]: The skill fetches the official
openaiPython package from the standard public registry (PyPI) at runtime. It also performs legitimate network operations to download generated video, thumbnail, and spritesheet assets from OpenAI's servers upon user request. - [CREDENTIALS_SAFE]: Security instructions explicitly forbid the agent from asking the user to provide API keys in the chat. Instead, it guides the user to set the
OPENAI_API_KEYas a local environment variable, adhering to secure credential management practices. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text to generate video prompts. It mitigates potential command injection and shell-escaping vulnerabilities by instructing the agent to use a file-based prompt input (
--prompt-file) and providing a structured template to wrap untrusted user data.
Audit Metadata