redline-docx-pl
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill invokes the 'adeu' tool using
uvxwith a strictly pinned version (adeu==1.30.0). This ensures that the agent uses a verified version of the dependency and prevents unexpected behavior from upstream updates. - [SAFE]: The skill is designed with a 'Safety Tier' framework, categorizing operations into Read-only, Mutating, and Destructive. Destructive operations (like permanent metadata removal) require the user to provide a specific verbal confirmation ('potwierdzam') before execution.
- [SAFE]: A core feature of the skill is 'sanitization,' which identifies and removes potentially sensitive metadata (author names, template paths, edit history) from DOCX files to ensure RODO/GDPR compliance when sharing documents externally.
- [SAFE]: Custom scripts such as
memo_negocjacyjne.pyandskan_placeholder.pyare implemented using the Python standard library with no external dependencies, minimizing the risk of third-party package vulnerabilities. - [SAFE]: The vendored Node.js engine (
vendor/docx-engine) is explicitly documented as inactive and is not integrated into the agent's execution path. It was used solely for measurement and benchmarking purposes during development.
Audit Metadata