narrator-ai-cli-skill
Warn
Audited by Gen Agent Trust Hub on Jun 15, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Installation of the core tool is recommended from a non-standard source (github.com/NarratorAI-Studio/narrator-ai-cli.git) instead of a verified package registry.
- [REMOTE_CODE_EXECUTION]: The pip installation from a remote Git URL involves the execution of setup scripts from an unverified source at the time of installation.
- [COMMAND_EXECUTION]: The skill instructions and provided Python examples use subprocess calls to execute shell commands, which can be misused if inputs are not sanitized.
- [DATA_EXFILTRATION]: The skill facilitates the uploading of user data and movie files to a remote API endpoint (api.narrator-ai.com) which is not among the whitelisted domains.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it passes user-controllable data to shell commands.
- Ingestion points: User-provided movie IDs, template selections, and categories in SKILL.md.
- Boundary markers: None identified in the provided instructions.
- Capability inventory: Shell command execution via CLI and subprocess.run.
- Sanitization: No explicit input sanitization or validation is implemented before command execution.
Audit Metadata