The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires installing a Python package directly from an unverified GitHub repository (https://github.com/NarratorAI-Studio/narrator-ai-cli.git) using pip install. This practice involves executing code from a third-party source that is not verified by standard package registries.
[REMOTE_CODE_EXECUTION]: The requirement to download and execute code from a remote, unverified repository constitutes a remote code execution risk, as the package content could be modified to include malicious payloads.
[COMMAND_EXECUTION]: The skill includes Python and Bash examples that utilize subprocess.run(..., shell=True) to execute commands. These scripts interpolate user-provided data, such as drama names, intros, and movie titles, directly into command strings without sanitization. This allows for command injection if the agent processes malicious input containing shell control characters.
[DYNAMIC_EXECUTION]: The skill uses local scripts to orchestrate a video generation pipeline, dynamically building and executing system commands based on output from the CLI tool.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user data that flows into high-privilege operations like shell execution.
Ingestion points: User-supplied movie names, drama titles, and descriptions found in the SKILL.md workflows.
Boundary markers: None; user data is embedded directly into shell command arguments without delimiters or warnings.
Capability inventory: The skill has access to the shell, filesystem, and network via the installed CLI tool.
Sanitization: There is no evidence of input validation, escaping, or filtering for the data interpolated into shell commands.

narrator-ai-cli-video-narration