Skip to content

Security: StevenGeller/torus

SECURITY.md

Security

Reporting Vulnerabilities

If you discover a security vulnerability, please report it privately via email to contact@steven-geller.com. Do not open a public issue.

I will acknowledge receipt within 48 hours and aim to release a fix within 7 days of confirmation.

Security Measures

Input Handling

  • Text input is sanitized and truncated (max 20 words) before processing
  • SVG output uses deterministic path generation with no user-controlled attributes
  • Uploaded SVGs for decode are parsed with strict path-data extraction only
  • No shell commands, SQL, or template interpolation on user input

File Uploads

  • The /api/decode endpoint accepts multipart file uploads
  • Only SVG path data (d attribute) is extracted; all other content is discarded
  • No uploaded files are written to disk or served back

Logging

  • Structured logging via tracing with RUST_LOG env filter (default: info)
  • access.log records timestamp, IP, and input text for operational visibility (capped at 5MB)
  • Log files are local only and not exposed via any endpoint

OG Image Cache

  • In-memory cache limited to 500 entries to prevent unbounded memory growth
  • Cache keys are derived from user input text; no filesystem writes

There aren't any published security advisories