[P2] Audit: Markdown, HTML, and SVG preview tools need explicit sanitization tests

[baixiangcpp]

Imported from the byteflow.tools deep audit on 2026-06-22.

Problem: tools that preview or transform user-controlled markup need explicit test coverage for safe rendering. Markdown Preview, HTML to Markdown, SVG Optimizer, SVG to PNG Converter, Open Graph preview, and similar tools should not rely only on manual review.

Affected tools:

• Markdown Preview
• HTML to Markdown
• SVG Optimizer
• SVG to PNG Converter
• Open Graph Meta Generator
• Any tool rendering user-provided HTML/SVG/metadata

Reproduction steps:

1. Open markup preview tools.
2. Test harmless inputs that mimic scripts, event handlers, iframes, external references, and SVG foreign content.
3. Verify whether content is sanitized, escaped, sandboxed, or rejected.
4. Check the CSP and console output.

Expected behavior: unsafe markup should not execute. Rendering should be sanitized, sandboxed, or safely escaped, and tests should enforce this.

Actual behavior: Trust Center mentions sanitization expectations, but explicit automated coverage needs confirmation.

Impact: user-content rendering risks can undermine trust and create security issues if a future change relaxes sanitization.

Proposed fix:

• Add a sanitizer test suite for markdown, HTML, SVG, and OG preview components.
• Add regression tests for script-like input, event attributes, external references, iframe-like content, and SVG special cases.
• Enforce CSP compatibility.
• Document the sanitization boundary per tool.

Acceptance criteria:

• Unsafe user-controlled markup never executes in tests.
• Sanitized output is predictable and visible to users.
• CSP does not need unsafe relaxations for previews.
• New preview tools must include sanitization tests before launch.

[baixiangcpp]

Closed as completed by the canonical BF audit backlog work. This older non-BF audit issue duplicates work now merged through PRs #209-#219, with the BF issues closed and guard/test coverage in place. The remaining non-duplicate follow-up is #7, which stays open because the upstream Next/PostCSS advisory is not fixed yet.