The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/init_skill.py performs a permission change using chmod(0o755) on a dynamically generated template script. This action allows the file to be executed as a program. While standard for a development tool, it represents a command execution capability on generated files.
[PROMPT_INJECTION]: The skill acts as a generator for other skills, which introduces a surface area for indirect prompt injection. Untrusted data used during the initialization or packaging process could influence the content of the resulting skill files.
Ingestion points: Arguments provided to scripts/init_skill.py (name and path).
Boundary markers: Not utilized in the generated templates to isolate user-provided content.
Capability inventory: File system writing, directory creation, and zip archiving.
Sanitization: Limited to basic naming regex and description checks for specific characters like angle brackets.

skill-creator