The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands like mkdir and project scripts like npm run validate to create the file structure and verify the generated SKILL.md. These are standard operations for a development utility.
[PROMPT_INJECTION]: The skill takes user input (name, category, workflow) and interpolates it into a markdown template. This creates a surface for indirect prompt injection, as user-provided data is placed into placeholders within the generated file.
Ingestion points: User-provided requirements gathered in the initial workflow step.
Boundary markers: The template uses [PLACEHOLDER] tokens to delimit user content.
Capability inventory: Uses mkdir for directory creation and npm for validation and testing.
Sanitization: Input strings are interpolated into the template without explicit sanitization or validation.

skill-creator