All skills
sugarforever avatar

/subtitle-correction

@1a5d219

Correct subtitle files (.srt) generated from speech recognition. Use when the user uploads subtitle files and asks to correct, fix, or proofread subtitles, especially for technical content like programming tutorials, AI/ML courses, or any content with domain-specific terminology. Supports Chinese and English subtitles with intelligent error detection and correction while preserving exact timeline information.

Use this Skill: https://skilld.dev/gh/sugarforever/01coder-agent-skills/subtitle-correction

This session only. Nothing lands on disk.

referencessrt-format.md

≈525 tokens on demand. Your agent reads this file only when SKILL.md points to it.

SRT Format Reference

File Structure

An SRT (SubRip Subtitle) file consists of sequential subtitle entries, each containing:

1
00:00:01,500 --> 00:00:04,500
First line of subtitle text
Optional second line

2
00:00:04,700 --> 00:00:07,666
Next subtitle text

Entry Components

1. Sequence Number

  • Sequential integer starting from 1
  • Must be preserved exactly
  • No gaps or duplicates allowed

2. Timestamp Line

HH:MM:SS,mmm --> HH:MM:SS,mmm
  • Start time --> End time
  • Hours:Minutes:Seconds,Milliseconds
  • Uses comma (,) as decimal separator, NOT period
  • Must be preserved exactly - NEVER modify

3. Text Content

  • One or more lines of text
  • This is the ONLY part to correct
  • Empty line separates entries

Critical Rules for Subtitle Correction

DO

  • Correct spelling and terminology errors in text lines
  • Maintain all line breaks within subtitle text
  • Preserve empty lines between entries
  • Keep sequence numbers unchanged
  • Keep timestamps unchanged

DON'T

  • Modify timestamp values
  • Change sequence numbers
  • Merge multiple entries
  • Split single entries
  • Add or remove entries
  • Change the comma in timestamps to period

Example Correction

Before (with errors):

53
00:02:23,433 --> 00:02:25,833
定义了一款工具summers conversation

54
00:02:26,133 --> 00:02:28,166
这款工具除了用了Lantern

After (corrected):

53
00:02:23,433 --> 00:02:25,833
定义了一款工具summarize_conversation

54
00:02:26,133 --> 00:02:28,166
这款工具除了用了LangChain

Processing Large Files

For files with many entries:

  1. Read in chunks using line ranges
  2. Track position to ensure complete coverage
  3. Verify entry count matches after correction
  4. Check continuity - no missing sequence numbers

Validation Checklist

Before outputting corrected file:

  • Entry count matches original
  • All sequence numbers present (1 to N)
  • All timestamps unchanged
  • No malformed timestamp lines
  • Empty line between each entry
  • No trailing whitespace issues

Source: SKILL.md on GitHub

1 warning14d5 checks · Risk SAFE
  • Gen Agent Trust Hub14d

    The skill is designed to correct speech recognition errors in subtitle files. It includes a local Python script for validation and diffing. While it processes untrusted user data, it follows strict structural rules and sanitizes output in its reporting tool.

  • Socket14d

    No alerts

  • Snyk14d

    Risk: LOW · No issues

  • Runlayer7mo

    4/4 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 1a5d219. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Steadyupdated 9 months ago

README badge

README badge for sugarforever/01coder-agent-skills/subtitle-correction