The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of a software package (narrator-ai-cli) directly from a GitHub repository (NarratorAI-Studio/narrator-ai-cli.git) rather than a verified package registry or a trusted vendor's repository.
[REMOTE_CODE_EXECUTION]: Installing code from a remote Git repository allows for arbitrary code execution on the host system during the installation process (via setup scripts) and whenever the command-line tool is invoked by the agent.
[COMMAND_EXECUTION]: The skill uses subprocess.run to execute commands on the local operating system, specifically for interacting with the narrator-ai-cli tool and managing local video and audio files.
[CREDENTIALS_UNSAFE]: The skill includes the narrator-ai-cli config show command, which displays the configured API key in plain text, potentially exposing sensitive credentials in logs or to the AI agent's context.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources and user requirements which could influence the behavior of the video generation process.
Ingestion points: External API data retrieved via movie list and template list, and user-provided strings passed to the --requirement flag in SKILL.md.
Boundary markers: Absent in the command patterns provided in SKILL.md.
Capability inventory: Ability to execute shell commands (subprocess.run), upload local files to a remote service, and download results via requests.get as described in SKILL.md.
Sanitization: No sanitization or validation of input requirements is specified before they are passed to the narration script generation process.

narrator-ai-video-generation