The Agent Skills Directory

[SAFE]: The skill is a pure-prompt instruction set for text editing. It does not utilize any scripts, executables, or external tools, eliminating the risk of remote code execution or system compromise.
[SAFE]: No evidence of data exfiltration or credential theft was found. The skill does not access sensitive file paths or perform network operations.
[PROMPT_INJECTION]: The skill instructions define a 'STYLE GUIDE OVERRIDE' where user-provided input can supersede internal principles. This is a standard functional requirement for editorial tools and does not constitute a malicious bypass of safety guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes external content and style_guide inputs without explicit boundary markers (e.g., XML tags or delimiters).
Ingestion points: content and style_guide input variables in SKILL.md.
Boundary markers: Absent; inputs are processed as raw text.
Capability inventory: None; the skill is limited to generating text output (a markdown table) and has no access to shell, network, or file system tools.
Sanitization: Absent; the skill performs only basic length and type validation on inputs.

bmad-editorial-review-prose