YouTube Transcript (yt-dlp)

Extract YouTube video transcripts from existing captions (manual or auto-generated) using yt-dlp, with optional timestamps and local SQLite caching. Use when...

Security Scan

The skill is coherent with its stated purpose (extracting YouTube transcripts via yt-dlp); required binaries and caching behavior make sense, but note the optional cookie handling and a small mismatch between documented env/config usage and the registry metadata.

Purpose & Capability

Name/description match what the included script and instructions do: use yt-dlp + python to download existing captions, parse VTT, and cache results in SQLite. Declared required binaries (python3, yt-dlp) are appropriate and necessary.

Instruction Scope

SKILL.md instructs the agent/user to run the bundled Python script, use a local SQLite cache under the skill baseDir, and optionally provide a Netscape-format cookies.txt. It also references an environment variable (YT_TRANSCRIPT_COOKIES) and a config path (~/.config/yt-transcript/) for cookies; these are operationally reasonable but the env/config usage is not declared in the registry metadata.

Install Mechanism

No install spec (instruction-only with a bundled script). This is low-risk; nothing in the manifest downloads arbitrary external code. The script runs yt-dlp as a subprocess, which is expected for this task.

Credentials

The skill does not require credentials in the registry metadata. However SKILL.md (and likely the script) optionally reads a cookies file and refers to YT_TRANSCRIPT_COOKIES. Cookies can contain sensitive session data and grant access to age-restricted or member-only content — treat them as secrets. The registry should have declared this optional env/config usage for transparency.

Persistence & Privilege

Skill is not always-enabled and is user-invocable. It writes a cache database under the skill baseDir (or user-specified path) which is appropriate for its functionality and not excessive privilege.

Assessment

This skill appears to do what it claims: it runs a bundled Python script that calls yt-dlp to fetch captions, parses VTT, and caches results locally. Before installing: (1) ensure you trust the yt-dlp binary on your system, (2) be aware the script writes a SQLite cache under the skill directory (or a custom path you provide), (3) if you need age-restricted content you must supply YouTube cookies — treat those cookie files (or YT_TRANSCRIPT_COOKIES env var) as sensitive and store them securely, and note that the registry metadata did not declare this optional env/config usage, (4) inspect the bundled script if you want to confirm it meets your privacy/security expectations. Overall coherent and proportionate for its purpose.

YouTube Transcript (Captions-Only)

Extracts transcripts from existing YouTube captions using yt-dlp. Prefers manual subtitles; falls back to auto-generated captions.

Prerequisites

• Python 3.7+
• yt-dlp installed and on PATH (pip install yt-dlp or system package)

How to Run

Script path: {baseDir}/scripts/yt_transcript.py

# Basic usage
python {baseDir}/scripts/yt_transcript.py <youtube_url_or_id>

# Specify language
python {baseDir}/scripts/yt_transcript.py <url> --lang en

# Plain text output
python {baseDir}/scripts/yt_transcript.py <url> --text

# Text without timestamps
python {baseDir}/scripts/yt_transcript.py <url> --text --no-ts

# Custom cache path
python {baseDir}/scripts/yt_transcript.py <url> --cache /path/to/cache.sqlite

Output Formats

JSON mode (default)

Returns a JSON object:

{
  "video_id": "dQw4w9WgXcQ",
  "lang": "en",
  "source": "manual",
  "segments": [
    { "start": 0.0, "duration": 4.2, "text": "We're no strangers to love" }
  ]
}

Text mode (--text)

Newline-separated transcript lines. Use --no-ts to omit timestamps.

Caching

Results are cached in a local SQLite database: {baseDir}/cache/transcripts.sqlite

Subsequent calls for the same video/lang/format are served from cache instantly.

To use a custom cache location: --cache /path/to/transcripts.sqlite

Cookies (optional)

For age-restricted or members-only videos, provide a Netscape-format cookies.txt:

export YT_TRANSCRIPT_COOKIES=/path/to/cookies.txt
python {baseDir}/scripts/yt_transcript.py <url>
# or
python {baseDir}/scripts/yt_transcript.py <url> --cookies /path/to/cookies.txt

Cookies must be stored under ~/.config/yt-transcript/ for security.

Troubleshooting

• No captions available: Video has no manual or auto-generated captions
• yt-dlp not found: Install with pip install yt-dlp or brew install yt-dlp
• Age-restricted video: Provide cookies from a logged-in YouTube session
• Rate limited: Wait and retry; reduce request frequency