The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides templates for executing CLI commands (e.g., novelai-image-mcp generate -p "...") that interpolate user-provided strings directly into shell environments.
[INDIRECT_PROMPT_INJECTION]: The skill defines pipelines that process untrusted user input (prompts) and pass them to external APIs and shell commands.
Ingestion points: User-provided text prompts used for image generation and transformation tools (e.g., prompt argument in generate_image).
Boundary markers: No delimiters or "ignore embedded instructions" warnings are used when interpolating user prompts into tool calls.
Capability inventory: The skill utilizes shell command execution, file system access for saving images, and network operations to interface with NovelAI APIs.
Sanitization: No input validation or shell escaping is specified for the prompt strings within the provided workflow templates.
[EXTERNAL_DOWNLOADS]: The skill mentions using uvx to execute novelai-image-mcp, which involves downloading and running the vendor's package from an external registry.

novelai-workflows