# Changelog of YouTube Transcript Scraper - Channels & Playlists (`datagrit/youtube-channel-playlist-transcripts`) Actor

- **URL**: https://apify.com/datagrit/youtube-channel-playlist-transcripts/changelog.md
- **Full Actor documentation**: https://apify.com/datagrit/youtube-channel-playlist-transcripts.md

### 0.4 (2026-10-01)

- Only new since last run on channels: videos in a temporary state are no longer remembered as read, so they do not become the stop point and are delivered by a later run: upcoming premieres and scheduled live streams, empty caption files, and videos without a publish date while a date window is set (bot checks and failed requests were already excluded).
- Field description and FAQ: on a channel these videos are read again only while they are newer than the stop point.

### 0.3 (2026-10-01)

- Only new since last run on channels: the Actor now remembers every video a run read on a channel (also those without a transcript: no captions, other language, wrong caption type, outside the filters) and stops the next listing at the first of them. Before, a video without a transcript between delivered ones let the listing continue into the older backlog, which was delivered and charged. Blocked videos and failed requests are not remembered.
- README FAQ: a video made public later with an older publish date can be missed by "only new".

### 0.2 (2026-10-01)

- Only new since last run: on channel uploads the listing now stops at the first already delivered videos, so scheduled runs return only new uploads instead of working through the older backlog; playlists and single videos still skip delivered videos.
- Videos cut by your language or caption-type choice no longer count as "videos with captions": such runs end green with free rows instead of failing with proxy advice. Status counter renamed to "videos with a matching caption track".
- A retired YouTube app client now fails the run with "the Actor needs an update" (rows with reason error) instead of proxy advice.
- Status warning with the count when more than half of the videos were blocked by a bot check.

### 0.1 (2026-10-01)

- First version: transcripts from any mix of videos, channels (@handle, /channel, /c, /user, /shorts and /streams tabs) and playlists.
- Per-channel limit, publish-date window (date or period), title keywords, channel content (long-form, Shorts, live replays, all).
- Caption language priority with optional fallback, uploaded vs auto-generated caption choice.
- Full text, timestamped segments and video metadata (publish date, duration, views, likes, category).
- Only-new-since-last-run memory per source and settings, with an optional memory name.
- Free rows with a reason for videos without a transcript; proxy session rotation on YouTube bot checks; the run fails when videos have captions but no transcript could be read.