The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script scripts/scaffold.py using the Bash tool to automate the creation of project structures and boilerplate code.
[EXTERNAL_DOWNLOADS]: Generated templates for README.md include instructions for users to download and install Python dependencies via pip, which is standard behavior for development-oriented skills.
[PROMPT_INJECTION]: The SKILL.md contains detailed instructions that define the agent's role as an 'Agent Skill Architect' and mandate specific interaction patterns, such as the 'Negative Instruction Constraint', to control model behavior during the design process.
[INDIRECT_PROMPT_INJECTION]: The skill represents an indirect injection surface because it ingests untrusted user data (such as skill descriptions) and interpolates them into generated markdown and script files.
Ingestion points: User-provided --name, --desc, and --path arguments.
Boundary markers: Generated files use YAML frontmatter and Markdown headers to delimit sections.
Capability inventory: The skill utilizes Bash for command execution and Write for file system modifications.
Sanitization: The name parameter is sanitized via regex (^[a-z0-9-]+$) in scaffold.py, although the description is directly formatted into templates without escaping.

create-skill