YouTube Full Channel Transcripts

Extract transcripts from all videos in a YouTube channel for free (no paid APIs). Uses yt-dlp to discover videos and fetch available subtitles. Saves combine...

Security Scan

The skill generally does what it says (uses yt-dlp and jq to fetch and parse subtitles), but there are small inconsistencies in requirements and declared environment variables that don't match the implementation.

Purpose & Capability

Name/description match the code: the script discovers videos with yt-dlp, downloads subtitles, converts SRT to text, and writes JSON/CSV. Required binaries (yt-dlp, jq) make sense. However the script invokes yt-dlp with '--js-runtimes node' but 'node' is not listed as a required binary—this is an actionable mismatch. Also skill.json declares environment vars (YT_DLP_MAX_RETRIES, YT_DLP_SLEEP_INTERVAL) that the script does not actually read.

Instruction Scope

Runtime instructions and the script operate within the stated scope: they call yt-dlp to fetch YouTube data, parse local files in /tmp, and write transcript outputs to an output directory. The script does not read unrelated system files or transmit data to third-party endpoints beyond YouTube/yt-dlp.

Install Mechanism

No install spec (instruction-only plus a shell script). Nothing is downloaded from arbitrary URLs; the script relies on system-installed binaries. This is low-risk from an installer perspective.

Credentials

The skill requests no credentials and declares only non-secret env vars in skill.json. But those declared vars (YT_DLP_MAX_RETRIES, YT_DLP_SLEEP_INTERVAL) are not actually used in the script, and the script expects input via lowercase variables like 'channel_url' which is not documented as an environment variable in skill.json. No sensitive credentials are requested.

Persistence & Privilege

The skill is not always-enabled and does not request elevated persistence. It does write temporary files to /tmp and outputs to a local output directory, which is expected behaviour for this utility.

What to consider before installing

This skill appears to implement the advertised functionality but has small inconsistencies you should be aware of before installing: 1) The script calls yt-dlp with '--js-runtimes node' but 'node' is not listed as a required binary—install Node.js or remove that option if you rely on yt-dlp's default. 2) skill.json declares YT_DLP_MAX_RETRIES and YT_DLP_SLEEP_INTERVAL, but the script does not use them; retries/backoff mentioned in README are not implemented. 3) The script expects input via environment variables named in lowercase (e.g., channel_url); the SKILL.md usage (prefixing the command with channel_url=...) will work but confirm how your agent passes parameters. 4) The script writes temporary files to /tmp and output files to a relative output_dir—run it in an isolated workspace if you need to protect local data. If these mismatches worry you, request an updated skill that: documents required 'node' if needed, actually uses or removes the declared env vars, and clarifies how parameters should be provided.

Free, local alternative to Apify actor — no third-party paid APIs.

Fetches all videos from a YouTube channel (or playlist), extracts available transcripts (auto-generated or manual), and exports to JSON/CSV.

When to Use

• "Get all transcripts from channel X"
• "Export channel Y's video subtitles"
• "Build dataset of all videos from creator Z"

Prerequisites

• yt-dlp installed (https://github.com/yt-dlp/yt-dlp)
• jq installed (for JSON processing)
• Internet connection

Usage

# Basic: fetch all transcripts from a channel
youtube_full_channel_transcripts channel_url="https://www.youtube.com/@SomeChannel" output_format="json"

# With filters
youtube_full_channel_transcripts channel_url="https://www.youtube.com/c/ChannelName" max_videos=50 languages="en" include_auto_generated=true

Options

Parameter	Type	Required	Description
channel_url	string	yes	YouTube channel URL, handle, or playlist URL
output_format	string	no	Output format: json (default) or csv
max_videos	integer	no	Limit number of videos to process (default: all)
languages	string	no	Comma-separated language codes (e.g., en,es). Default: en
include_auto_generated	boolean	no	Include auto-generated transcripts? (default: true)
output_dir	string	no	Directory to save results (default: workspace/exports/youtube-transcripts)

Output

• transcripts.json or transcripts.csv — structured data with:

  {
    "video_id": "abc123",
    "title": "Video Title",
    "url": "https://youtube.com/watch?v=abc123",
    "upload_date": "20240322",
    "duration": 360,
    "language": "en",
    "transcript": "Full text...",
    "is_auto_generated": false
  }
  

• Summary printed to console: number of videos processed, success/failure counts.

Implementation Details

1. Discovery: yt-dlp --flat-playlist -J to list all videos
2. Transcript fetch: yt-dlp --write-subs --sub-format srt --skip-download per video
3. Parsing: Extract SRT subtitles and convert to plain text
4. Combine: Aggregate results into JSON/CSV

Errors & Retries

• If a video has no available transcript, it's skipped (logged)
• Network errors: retry up to 3 times with backoff
• Rate limiting: yt-dlp handles automatically; can add --sleep-interval if needed

Notes

• Respect YouTube's rate limits. The tool includes small delays by default.
• Large channels (100+ videos) may take a while; use max_videos to limit.
• Auto-generated transcripts are more available but less accurate.