The Agent Skills Directory

[COMMAND_EXECUTION]: The skill environment executes a local JavaScript grader (graders/check-skill.js) to verify the structure of skill files. This script performs read operations on local workspace files.
[EXTERNAL_DOWNLOADS]: Metadata references external resources including agentskills.io and a public GitHub repository mgechev/skillgrade. These are documented as references for the skill's domain of skill authoring.
[PROMPT_INJECTION]: The skill acts as an editor for other agent skills, processing draft instruction files. This creates a surface for indirect prompt injection, where content within a processed draft could attempt to manipulate the agent's behavior. This is an inherent risk of instruction-editing tools and is managed by standard agent safety protocols.

skill-creator