Transparency

Testing methodology and known limitations

This page explains what is checked before publishing a tool, what the tests do not prove, and how users can verify important behavior independently.

Local-processing checks

Tools are tested with synthetic values while the browser Network panel is monitored. The expected behavior is that pasted content and selected file bytes do not appear in requests to a Privacy Toolbox processing API.

Functional test cases

Text tools are checked against representative emails, phone numbers, IDs, URLs, tokens, JSON nesting, CSV quoting, and JWT payloads. Image tools are checked with supported formats, orientation, dimensions, export, and common metadata fields.

Regression and build checks

The project validates data references, duplicate slugs, internal links, sitemap URLs, JavaScript syntax, and production output. Important redirects are kept in version-controlled configuration.

What these tests do not prove

Tests do not guarantee complete anonymization, forensic metadata removal, regulatory compliance, security against malicious browser extensions, or compatibility with every file produced by every application.

  • Pattern detection can produce false positives and false negatives
  • Lightweight metadata parsing does not cover every proprietary field
  • Browser memory limits vary by device
  • High-risk workflows require approved professional tools

Imran Gul

Web developer and maintainer of Privacy Toolbox. Tool claims are intentionally limited to the implemented browser workflows and representative tests described on this site.

Relevant utilities

Recommended tools

Review recent changes

See what was added, consolidated, fixed, and clarified.

Continue