The Agent Skills Directory

[SAFE]: The skill consists of a comprehensive set of Markdown-based guidelines for prompt design and auditing. It does not contain any executable code, scripts, or binary files.
[DATA_EXPOSURE]: Analysis confirms the skill only references its own internal documentation files within the ${SKILL_DIR}/docs/ directory. There are no attempts to access sensitive system files, environment variables, or hardcoded credentials.
[PROMPT_INJECTION]: The skill's instructions define specific roles for auditing and diagnosis. These are functional role assignments rather than malicious overrides or bypass attempts.
[PROMPT_INJECTION]: The skill analyzes user-provided prompt text, creating an indirect prompt injection surface. Ingestion points: docs/workflow-audit.md and docs/workflow-diagnose.md. Boundary markers: Functional instructions to treat the target prompt as data for analysis. Capability inventory: Accessing internal documentation and generating reports. Sanitization: Not present, but the skill lacks dangerous tools to exploit potential injections.

prompt-helper