video-chapters · Video Chapter Index
Video chapter / timestamp index generator: for medium and long videos (tutorials, reviews, lectures, stream replays), it segments chapters and produces a timestamp directory ready to paste into Bilibili/YouTube descriptions, so viewers can jump to what they want.
First transcribes with timestamped ASR, then splits chapters at topic transitions and names them: chapter one must start at 00:00, adjacent chapters ≥10s apart, names highlight the payoff. Outputs chapters.txt and structured chapters.json.
Example invocation: "Generate a chapter index for this 40-minute tutorial video."
Full brief
Positioning
video-chapters is a video chapter timestamp generator: it outputs only the chapter index — no full-article rewrite, no subtitles. It solves one need: letting viewers jump to the parts they care about.
Core capabilities
- Timestamped transcription: faster-whisper (asr.py) produces per-segment start/end transcripts as the basis for chaptering.
- Topic-transition splitting: chapter boundaries follow subject changes, not even time slices; typically 5–12 chapters; shorts under 3 minutes usually need none.
- Chapter naming: 6–16 characters each, verb-led or payoff-pointing — never empty labels like "Part 1".
- Platform-rule compliance: first chapter must be 00:00 (or the platform's chapter feature won't activate); adjacent chapters ≥10s apart (YouTube hard rule).
- Dual-format output:
chapters.txt(onemm:ss titleper line, paste-ready) +chapters.json(start seconds + title, for programmatic use).
Workflow
- Timestamped ASR transcription of the video
- Split chapters at topic transitions and name them (faithful to the transcript; nothing invented)
- Output
chapters.txt,chapters.json, plus the raw transcript for reference
Inputs & outputs
| Input | Notes |
|---|---|
| Video file | Required — medium/long video (tutorial / review / lecture / stream replay) |
| Target platform | Optional — Bilibili / YouTube / generic (affects format and wording) |
| Chapter count | Optional — desired count (defaults to natural content splits) |
Output: chapters.txt, chapters.json, transcript.txt under outputs/<theme>/.
Boundaries with adjacent skills
| Skill | Lane |
|---|---|
| video-chapters (this) | Chapter timestamp index only |
| video-to-article | Rewrites the video into a full article |
| auto-subtitle | Generates subtitle files |
| video-highlights | Cuts long videos into short clips |
Fit
- Chapter navigation in Bilibili/YouTube long-video descriptions
- Payoff indexes for courses, lectures, stream replays
- Projects needing a structured timeline for downstream development
Before you start
- First ASR run needs a proxy to download the model (see auto-subtitle prerequisites).
- Target platform affects format details; chapters follow the transcript — no invented content.
video-chapters is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.