natural-korean
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes local Python utility scripts (
scripts/lint.py,scripts/outline.py,scripts/terms.py) via theuv runcommand to perform linguistic analysis on documents.- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided Korean documents for rewriting and style correction, creating a potential attack surface for indirect prompt injection. While the skill includes instructions to the agent to treat findings as suggestions, it lacks formal input sanitization. - Ingestion points: Document files provided by the user for writing, correction, or diagnosis as described in
SKILL.md(Step 4 and 6). - Boundary markers: Absent. The prompt instructions do not define structural delimiters to isolate user-supplied content from the agent's instructions.
- Capability inventory: The agent executes shell commands (
uv run) and performs local file read/write operations to manage linting results. - Sanitization: Absent. No explicit filtering or sanitization of the input document content is performed prior to processing.
Audit Metadata