The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard shell commands such as mkdir, cp, and git to manage the directory structure and versioning of skills within the personal AI environment.
[EXTERNAL_DOWNLOADS]: References the official Anthropic skills repository on GitHub (github.com/anthropics/skills) for learning and examples. Anthropic is a well-known and trusted organization.
[PROMPT_INJECTION]: The skill processes user input to generate new instruction files, creating an attack surface for indirect prompt injection.
Ingestion points: User descriptions and logic provided during the skill creation workflow in workflows/create-new.md.
Boundary markers: Templates provide structure but do not explicitly wrap user input in delimiters to prevent instructions from being interpreted as commands by the agent when the generated skill is later used.
Capability inventory: File creation, directory management, and git operations.
Sanitization: No explicit sanitization or validation of user-provided strings is documented in the templates.

create-skill