narrator-ai-cli-skill
Fail
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill installs a Python package directly from a GitHub repository using
pip install "narrator-ai-cli @ git+https://github.com/NarratorAI-Studio/narrator-ai-cli.git". This allows for the download and execution of arbitrary code from an external, unverified source during the installation phase.\n- [EXTERNAL_DOWNLOADS]: The skill usescurlto download video files from URLs provided by the API (VIDEO_URL). These URLs are generated at runtime and point to external servers.\n- [COMMAND_EXECUTION]: The skill relies onsubprocess.runto call thenarrator-ai-clitool with various arguments, including file IDs and movie IDs, which may be influenced by external input.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, specifically movie files andclips.jsonmetadata, to generate narration scripts.\n - Ingestion points: Movie files and JSON clip metadata uploaded via
narrator-ai-cli file upload.\n - Boundary markers: None identified; the skill directly passes file references to the CLI for processing.\n
- Capability inventory: The skill can execute shell commands via
subprocess.run, perform network operations viacurlandhttpx, and write files to the local system.\n - Sanitization: No evidence of input validation or sanitization for the content of the uploaded movies or JSON metadata before they are used to influence AI script generation.
Recommendations
- AI detected serious security threats
Audit Metadata