CaptionGate — SRT Quality Check

Catch subtitle problems before your next video handoff.

Give CaptionGate an existing UTF-8 SRT. It checks timestamp structure, duration, ordering, overlaps, numbering, reading speed and line length. Each finding points to the original source block and line.

Deliverables

• Standalone HTML report for visual review.
• Markdown and JSON reports for handoff and reuse.
• SHA256 evidence manifest.
• Optional format-normalized SRT: sequential numbering and consistent separators/newlines, preserving timestamp and text lines. Errors block this copy; overlap and readability warnings are not repaired.

Example

A synthetic 3-block SRT produces 0 errors and 8 warnings, including a 0.5-second overlap and a fast-reading cue. This is a demonstration, not a customer result.

Requirements and limits

Python 3.10+ and a skill-capable AI client with local file execution. Standard library only. No script API key or third-party packages. EN/ES assistant explanations; generated reports are English. UTF-8 SRT, up to 10 MiB and 20,000 blocks. Editable heuristics are not platform standards.

No transcription, translation, audio alignment, retiming, video rendering or publishing. Always preview with the actual video. No accuracy, accessibility, compliance or approval guarantee.

Data and license

The Python script has no network calls or telemetry. Reports contain source captions. Your AI client may process uploads under its policies; the complete workflow is not necessarily offline. Publisher operates no backend. Single-buyer internal/client use and commercial outputs allowed; redistribution/resale of the skill prohibited. AI-client fees are separate. No support SLA.