The Agent Skills Directory

[SAFE]: The skill is a meta-process tool designed to improve the reliability of other skills through prompt engineering patterns like 'Iron Laws' and 'Gate Functions'.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes untrusted user input (intent capture) to generate new instructions.
Ingestion points: User intent capture and interview process in Step 2 of the SKILL.md file.
Boundary markers: Absent; the skill relies on standard markdown delimiters.
Capability inventory: Invokes the Skill() tool for delegation and the Read() tool for local file access.
Sanitization: No explicit input sanitization or validation logic is defined to filter malicious prompts from the generated skill drafts.
[DATA_EXFILTRATION]: No evidence of data exposure or exfiltration. File access is confined to internal reference documentation (e.g., enforcement-checklist.md) within the plugin environment, and no external network operations are present.
[COMMAND_EXECUTION]: The skill does not execute shell commands or binary files. All operations are performed through high-level agentic tool invocations.

skill-creator