polish-language
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses bundled Python scripts for deterministic manuscript linting. The test suite includes
subprocess.runcalls to execute these scripts against synthetic fixtures, ensuring logic consistency across US and UK spelling variants. - [DYNAMIC_EXECUTION]: Test files utilize
importlibto dynamically load internal linter modules from relative file paths. This is limited to skill-internal resources for verification purposes. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes user-provided manuscript text (Markdown and plain text). While it implements deterministic reporting and requires explicit human approval for all edits, the ingestion of untrusted data constitutes a potential indirect injection surface.
- [SAFE]: Triggers include localized Korean text terms for "consistency check" and "proofreading," which are legitimate non-English functional descriptions and not indicators of homoglyph obfuscation.
Audit Metadata