tk-blind-comparison
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for offline human comparison of items. It generates self-contained HTML galleries that do not require network access for core functionality.
- [SAFE]: Artifacts are stored in a local
.tigerkit/directory. The skill includes instructions and scripts to ensure this directory is properly ignored by Git to prevent accidental data exposure. - [SAFE]: Credential management for local application authentication is handled through protected files with restricted permissions. The skill explicitly forbids these credentials from being included in version control or exported data.
- [SAFE]: The Python scripts (build_gallery.py, build_html.py, etc.) perform standard data processing and HTML assembly. They use established escaping methods to prevent injection vulnerabilities and do not perform unauthorized network requests.
- [SAFE]: Blinding of candidate identities is implemented using Base64-encoded mappings within the HTML. This is a functional requirement for blind testing and is not used to hide malicious behavior.
Audit Metadata