The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/init_skill.py performs filesystem management and permission modifications.
It creates directory structures using Path.mkdir based on a user-provided path and skill name.
It generates boilerplate markdown and Python files using Path.write_text.
It modifies file permissions of generated example scripts to set them as executable using os.chmod.
The script scripts/package_skill.py reads a directory tree and generates a compressed ZIP archive.
[PROMPT_INJECTION]: The skill exhibits an indirect injection surface where user-provided input is used in script generation.
Ingestion points: The skill_name argument in scripts/init_skill.py is interpolated directly into templates for SKILL.md and example.py.
Boundary markers: There are no boundary markers or delimiters around the interpolated skill_name to prevent it from containing instructions that might be interpreted by the agent during subsequent use of the generated skill.
Capability inventory: The skill allows for recursive filesystem writes and the creation of executable files.
Sanitization: No input sanitization is performed on the skill_name identifier before it is used to construct file paths and template contents in the initialization script.

skill-creator