The Agent Skills Directory

[SAFE]: The skill provides a structured framework for analyzing Korean documents for style consistency (tone, terminology, numbering, etc.) based on established government and academic standards.\n- [SAFE]: All external URLs included in the documentation (e.g., links to the National Institute of Korean Language and academic style guides) are for legitimate reference and point to well-known and trusted domains.\n- [SAFE]: Example credentials and tokens found in technical documentation examples (e.g., client_id=YOUR_KEY, access_token": "eyJ...") are clearly marked as placeholders and do not represent actual exposed secrets.\n- [PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection because it is designed to ingest and process untrusted user-provided text files using the Read tool and subsequently modify files using the Write and Edit tools. Ingestion point: SKILL.md Workflow Step 1. Capability inventory: Read, Write, Edit, Grep, Glob, AskUserQuestion. Boundary markers: Absent (instructions guide the agent to process the text directly). Sanitization: Absent. However, the skill instructions themselves are benign and focused solely on stylistic correction.

korean-style-guide