Forgiving input
Time-first, title-first, dashes, pipes, colons, brackets: mixed freely, line by line.
Chapter lists travel: the same ten lines go into a YouTube description, a video platform’s chapter field, a spreadsheet for review, and sometimes an API call. Each destination wants its own arrangement, and re-typing timestamps is exactly the kind of work that produces 1:02 where 1:20 was meant.
The formatter parses the input before it changes the format. It recognizes leading, trailing, and bracketed timestamps. It cleans the remaining text as the title, reports unreadable lines by number, and sorts the parsed list by time.
The YouTube output enforces YouTube’s actual rules: the list must begin at 0:00, contain at least 3 chapters, and give each chapter at least 10 seconds. Lists that break a rule still format, with a note saying which rule: because sometimes the list is for somewhere else.
The tool uses fixed local rules without AI or an upload. Use the AI video chapter generator if you want to create chapters from a transcript.