Communitygithub.com

daymade/claude-code-skills

Download YouTube videos and HLS streams (m3u8) from platforms like Mux, Vimeo, etc. using yt-dlp and ffmpeg. Use this skill when users request downloading videos, extracting audio, handling protected streams with authentication headers, or troubleshooting download issues like nsig extraction failures, 403 errors, or cookie extraction problems.

What is claude-code-skills?

claude-code-skills is a Claude Code agent skill that download YouTube videos and HLS streams (m3u8) from platforms like Mux, Vimeo, etc. using yt-dlp and ffmpeg. Use this skill when users request downloading videos, extracting audio, handling protected streams with authentication headers, or troubleshooting download issues like nsig extraction failures, 403 errors, or cookie extraction problems.

Works with✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/daymade/claude-code-skills/tree/HEAD/youtube-downloader

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Preview

From the skill README

Thumbnail
Thumbnail

Documentation

YouTube Downloader

Overview

Enable reliable video and audio downloads from YouTube and HLS streaming platforms (Mux, Vimeo, etc.) using yt-dlp and ffmpeg. This skill provides workflows for:

  • YouTube downloads (up to 4K) using PO token providers or browser cookies
  • HLS stream downloads with authentication headers
  • Handling protected content and troubleshooting common download failures

Non-Technical User Experience (Default)

Assume the user is non-technical. Do not ask them to run commands. Execute everything yourself and report progress in plain language. Avoid mentioning tooling unless the user asks.

Default flow:

  1. Ask for the URL (if not provided).
  2. Fetch video metadata (title/uploader/duration/thumbnail) and confirm it matches the user's intent.
    • If yt-dlp is blocked by “confirm you’re not a bot”, fall back to YouTube oEmbed for title/uploader/thumbnail (duration may be unknown).
  3. Offer simple choices (video vs. audio-only, quality, subtitles, save location).
  4. Proceed with sensible defaults if the user does not specify:
    • Video download at best quality
    • MP4 merged output
    • Single video only (no playlists)
  5. Download and report the final file path, file size, and resolution (if video).

Offer choices in user-friendly terms:

  • “Download the video in best quality (default)”
  • “Download audio only (MP3)”
  • “Pick a quality: 1080p / 720p / 480p / 360p”
  • “Include subtitles (if available)”
  • “Save to the Downloads folder (default) or tell me another folder”

Always render the thumbnail when available:

  • If metadata includes a thumbnail URL, include it using Markdown image syntax: ![Thumbnail](URL).

Ask before doing extra work:

  • Confirm playlist downloads (can be large).
  • Confirm installing/upgrading dependencies if missing.
  • Ask before extracting browser cookies.
  • If using cookies, never mention cookie counts or raw cookie details in user-facing responses. Say “used your Chrome login session”.
  • If verification is required, automatically set up a local PO Token helper (no user actions). If Docker is missing or fails, do not attempt to install Docker—switch to the browser-based PO Token provider instead.

Legal/Safety reminder (brief):

  • Proceed only if the user has the rights or permission to download the content.

Response template (use plain language, no commands):

![Thumbnail](THUMBNAIL_URL)

Title: …
Channel: …
Duration: …

I can help you:
1) Download the video (best quality, MP4)
2) Download audio only (MP3)
3) Pick a specific quality (1080p/720p/480p/360p)
4) Include subtitles (if available)

Where should I save it? (Default: Downloads folder)

If the user says “just download”:

  • Proceed with defaults and confirm when the download finishes.
    • If blocked by a 403, automatically set up the verification helper and retry.

Reliable Download SOP (Internal)

Follow this SOP to avoid common failures and confusion:

  1. Quote URLs in shell commands (zsh treats ? as a glob). Example: 'https://www.youtube.com/watch?v=VIDEO_ID'.
  2. Ensure proxy is active for both yt-dlp and PO Token providers (HTTP_PROXY/HTTPS_PROXY/ALL_PROXY).
  3. If you see “Sign in to confirm you’re not a bot”, request permission and use browser cookies. Do not proceed without cookies.
  4. Start a PO Token provider before downloading (fail fast if it cannot start).
    • Use Docker bgutil provider when available.
    • If Docker is missing or fails, switch to browser-based WPC provider.
  5. If cookies are in use, prefer the web_safari player client. Otherwise prefer mweb for PO tokens.
  6. Keep the browser window open while WPC is minting tokens. Ensure Chrome can reach YouTube through the same proxy.
  7. If you get “Only images are available” or “Requested format is not available”, treat it as a PO Token failure and retry after fixing token provider/browser state.
  8. If you get SSL EOF or fragment errors, treat it as a proxy/network issue. Retry with progressive formats and/or a better proxy.

Agent Execution Checklist (Internal)

  • Run scripts/download_video.py URL --info (add --cookies-from-browser chrome if permission granted) to fetch metadata and thumbnail.
  • If yt-dlp metadata fails, rely on the script’s oEmbed fallback for title/uploader/thumbnail and note that duration may be unavailable.
  • If a thumbnail URL is present, render it in the response with Markdown image syntax.
  • Ask the user to choose video vs. audio-only and (optionally) a quality preset.
  • Use a friendly default save location (Downloads folder) unless the user specifies a folder.
  • For subtitles, run with --subtitles and the requested --sub-lang.
  • After download, report file name, size, and resolution (if video) in plain language.
  • If download fails with 403/fragment errors, retry once with non-m3u8 progressive formats.
  • If “Sign in to confirm you’re not a bot” appears, request cookie access and retry with cookies + web_safari.
  • If “Only images are available” appears, treat it as PO Token failure and retry after fixing provider/browser state.
  • Start the PO Token provider before downloads (--auto-po-token default). Fail fast if it cannot start.
  • If Docker-based provider fails (common in China), automatically fall back to the browser-based WPC provider (it may briefly open a browser window).
  • If the WPC provider is used, keep the browser window open until download starts. If the browser fails to launch, set the Chrome path explicitly.
  • If the PO Token provider times out, restart it once and retry.
  • If a system proxy is configured, pass it into the provider container. If the proxy points to 127.0.0.1/localhost, rewrite it to host.docker.internal for Docker.

When to Use This Skill

This skill should be invoked when users:

  • Request downloading YouTube videos or playlists
  • Want to extract audio from YouTube videos
  • Experience yt-dlp download failures or limited format availability
  • Need help with format selection or quality options
  • Report only low-quality (360p) formats available
  • Ask about downloading YouTube content in specific quality (1080p, 4K, etc.)
  • Need to convert downloaded WebM videos to MP4 format for wider compatibility
  • Request downloading HLS streams (m3u8) from platforms like Mux, Vimeo, or other streaming services
  • Need to download protected streams that require authentication headers

Prerequisites

1. Verify yt-dlp Installation (Run this yourself)

which yt-dlp
yt-dlp --version

If not installed or outdated (< 2025.10.22):

brew upgrade yt-dlp  # macOS
# or
pip install --upgrade yt-dlp  # Cross-platform

Critical: Outdated yt-dlp versions cause nsig extraction failures and missing formats.

2. Check Current Quality Access (Run this yourself)

Before downloading, check available formats:

yt-dlp -F "https://youtu.be/VIDEO_ID"

If only format 18 (360p) appears: PO token provider setup needed for high-quality access.

High-Quality Download Workflow

Step 1: Install PO Token Provider (One-time Setup)

For 1080p/1440p/4K access, install a PO token provider plugin into yt-dlp's Python environment:

# Find yt-dlp's Python path (interpreter used by yt-dlp)
head -1 $(which yt-dlp)

# Install plugin using the interpreter from the line above
<YTDLP_PYTHON> -m pip install bgutil-ytdlp-pot-provider

Verification: Run yt-dlp -F "VIDEO_URL" again. Look for formats 137 (1080p), 271 (1440p), or 313 (4K).

See references/po-token-setup.md for detailed setup instructions and troubleshooting.

Step 2: Download with Best Quality

Once PO token provider is installed:

# Download best quality up to 1080p
yt-dlp -f "bestvideo[height<=1080]+bestaudio/best" "VIDEO_URL"

# Download best available quality (4K if available)
yt-dlp -f "bestvideo+bestaudio/best" "VIDEO_URL"

Step 3: Verify Download Quality

# Check video resolution
ffprobe -v error -select_streams v:0 -show_entries stream=width,height,codec_name -of default=noprint_wrappers=1 video.mp4

Expected output for 1080p:

codec_name=vp9
width=1920
height=1080

Alternative: Browser Cookies Method

If PO token provider setup is problematic, use browser cookies:

# Firefox
yt-dlp --cookies-from-browser firefox -f "bestvideo[height<=1080]+bestaudio/best" "VIDEO_URL"

# Chrome
yt-dlp --cookies-from-browser chrome -f "bestvideo[height<=1080]+bestaudio/best" "VIDEO_URL"

Benefits: Access to age-restricted and members-only content. Requirements:

  • Must be logged into YouTube in the specified browser.
  • Browser and yt-dlp must use the same IP/proxy.
  • Do not use Android client with cookies (Android client does not support cookies).

Common Tasks

Audio-Only Download (Run this yourself)

Extract audio as MP3:

yt-dlp -x --audio-format mp3 "VIDEO_URL"

Custom Output Directory (Run this yourself)

yt-dlp -P ~/Downloads/YouTube "VIDEO_URL"

Download with Subtitles (Run this yourself)

yt-dlp --write-subs --sub-lang en "VIDEO_URL"

Playlist Download (Run this yourself)

yt-dlp -f "bestvideo[height<=1080]+bestaudio/best" "PLAYLIST_URL"

Convert WebM to MP4 (Run this yourself)

YouTube high-quality downloads often use WebM format (VP9 codec). Convert to MP4 for wider compatibility:

# Check if ffmpeg is installed
which ffmpeg || brew install ffmpeg  # macOS

# Convert WebM to MP4 with good quality settings
ffmpeg -i "video.webm" -c:v libx264 -preset medium -crf 23 -c:a aac -b:a 128k "video.mp4"

Parameters explained:

  • -c:v libx264: Use H.264 video codec (widely compatible)
  • -preset medium: Balance between encoding speed and file size
  • -crf 23: Constant Rate Factor for quality (18-28 range, lower = better quality)
  • -c:a aac: Use AAC audio codec
  • -b:a 128k: Audio bitrate 128 kbps

Tip: Conversion maintains 1080p resolution and provides ~6x encoding speed on modern hardware.

Troubleshooting Quick Reference

Only 360p Available (Format 18)

Cause: Missing PO token provider or outdated yt-dlp.

Solution:

  1. Update yt-dlp: brew upgrade yt-dlp
  2. Install PO token provider (see Step 1 above)
  3. Or use browser cookies method

Sign in to Confirm You’re Not a Bot

Cause: YouTube requires authentication to proceed.

Solution:

  1. Request permission and use browser cookies (--cookies-from-browser chrome).
  2. Ensure the browser and yt-dlp use the same IP/proxy.
  3. Retry with web_safari client if needed.

Only Images Available / Requested Format Not Available

Cause: PO tokens not applied or provider/browser verification failed.

Solution:

  1. Verify PO Token provider is running before download.
  2. Keep the browser window open if using WPC.
  3. If cookies are in use, prefer web_safari client and retry.

nsig Extraction Failed

Symptoms:

WARNING: [youtube] nsig extraction failed: Some formats may be missing

Solution:

  1. Update yt-dlp to latest version
  2. Install PO token provider
  3. If still failing and PO tokens are disabled, use Android client: yt-dlp --extractor-args "youtube:player_client=android" "VIDEO_URL"

SSL EOF / Fragment Errors

Cause: Proxy or network instability.

Solution:

  1. Retry with progressive formats (non-m3u8).
  2. Switch to a more stable proxy/node.
  3. Avoid closing the PO token browser window during download.

Slow Downloads or Network Errors

For users in China or behind restrictive proxies:

  • Downloads may be slow due to network conditions
  • Allow sufficient time for completion
  • yt-dlp automatically retries on transient failures

PO Token Warning (Harmless)

WARNING: android client https formats require a GVS PO Token

Action: Ignore if download succeeds. This indicates Android client has limited format access without PO tokens.

Bundled Script Reference

scripts/download_video.py

Use this convenience wrapper to auto-start a PO Token provider by default for high-quality downloads. Use it yourself and report results to the user without asking them to run commands.

Basic usage:

scripts/download_video.py "VIDEO_URL"

Arguments:

  • url - YouTube video URL (required)
  • -o, --output-dir - Output directory
  • --output-template - Output filename template (yt-dlp syntax)
  • -f, --format - Format specification
  • -q, --quality - Quality preset (best, 1080p, 720p, 480p, 360p, worst). Default: best (skipped for --audio-only)
  • -a, --audio-only - Extract audio as MP3
  • --subtitles - Download subtitles if available
  • --sub-lang - Subtitle languages (comma-separated, default: en)
  • --cookies-from-browser - Load cookies from a browser (e.g., chrome, firefox)
  • --cookies-file - Load cookies from a cookies.txt file
  • --player-client - Use a specific YouTube player client (e.g., web_safari)
  • --auto-po-token - Auto-start PO Token provider (default; uses Docker if available, otherwise switches to browser-based provider)
  • --no-auto-po-token - Disable auto PO Token setup
  • --proxy - Proxy URL for yt-dlp and the PO Token provider (e.g., http://127.0.0.1:1082)
  • --wpc-browser-path - Browser executable path for WPC provider
  • -F, --list-formats - List available formats
  • --merge-format - Merge output container (e.g., mp4, mkv). Default: mp4
  • --playlist - Allow playlist downloads (default: single video only)
  • --info - Print title/uploader/duration/thumbnail and exit
  • --no-android-client - Disable Android client fallback

Note: Use the Android client only when PO tokens are disabled. Keep PO tokens enabled for high quality.

Quality Expectations

Setup360p720p1080p1440p4K
Auto PO token (default)✓✓✓✓✓
Android client only✓✗✗✗✗
PO token provider (manual)✓✓✓✓✓
Browser cookies✓✓✓✓✓

HLS Stream Downloads (m3u8)

For streaming platforms like Mux, Vimeo, and other HLS-based services, use ffmpeg as the primary tool. These streams often require authentication headers that yt-dlp may not handle correctly.

Identifying HLS Streams

HLS streams use .m3u8 playlist files:

  • Master playlist: Lists multiple quality options
  • Rendition playlist: Contains actual video/audio segment URLs

Download Workflow

Step 1: Obtain the Stream URL

Get the m3u8 URL from the video source. For protected streams:

  1. Open browser DevTools → Network tab
  2. Play the video
  3. Filter for "m3u8" to find the playlist URLs
  4. Copy the rendition URL (usually contains quality info like "rendition.m3u8")

Step 2: Identify Required Headers

Many CDNs require authentication headers:

  • Referer: Origin website (e.g., https://maven.com/)
  • Origin: Same as Referer for CORS
  • User-Agent: Browser identification

Check the Network tab to see which headers the browser sends.

Step 3: Download with ffmpeg

Use ffmpeg with the -headers flag for protected streams:

ffmpeg -headers "Referer: https://example.com/" \
  -protocol_whitelist file,http,https,tcp,tls,crypto,httpproxy \
  -i "https://cdn.example.com/path/rendition.m3u8?params" \
  -c copy -bsf:a aac_adtstoasc \
  output.mp4

Key parameters:

  • -headers: Set HTTP headers (critical for authentication)
  • -protocol_whitelist: Enable required protocols for HLS
  • -c copy: Stream copy (no re-encoding, faster)
  • -bsf:a aac_adtstoasc: Fix AAC audio compatibility

Common header patterns:

# Single header
-headers "Referer: https://example.com/"

# Multiple headers
-headers "Referer: https://example.com/" \
-headers "User-Agent: Mozilla/5.0..."

# Alternative syntax
-headers $'Referer: https://example.com/\r\nUser-Agent: Mozilla/5.0...'

Handling Separate Audio/Video Streams

Some platforms (like Mux) deliver audio and video separately:

  1. Download audio stream:
ffmpeg -headers "Referer: https://example.com/" \
  -protocol_whitelist file,http,https,tcp,tls,crypto,httpproxy \
  -i "https://cdn.example.com/audio/rendition.m3u8" \
  -c copy audio.m4a
  1. Download video stream:
ffmpeg -headers "Referer: https://example.com/" \
  -protocol_whitelist file,http,https,tcp,tls,crypto,httpproxy \
  -i "https://cdn.example.com/video/rendition.m3u8" \
  -c copy video.mp4
  1. Merge streams:
ffmpeg -i video.mp4 -i audio.m4a -c copy merged.mp4

Troubleshooting HLS Downloads

403 Forbidden Errors

Cause: Missing or incorrect authentication headers.

Solution:

  1. Verify Referer header matches the video source website
  2. Check if additional headers (Origin, User-Agent) are needed
  3. Ensure the m3u8 URL includes all query parameters from browser

yt-dlp Stuck on Cookie Extraction

Symptom: Extracting cookies from chrome hangs indefinitely.

Solution: Use ffmpeg directly instead of yt-dlp for HLS streams.

Protocol Not Whitelisted

Error: Protocol 'https' not on whitelist 'file,crypto,data'

Solution: Add -protocol_whitelist file,http,https,tcp,tls,crypto,httpproxy

Empty Segments or No Streams

Cause: Expired signatures in the m3u8 URLs.

Solution:

  1. Get fresh URLs from browser DevTools
  2. Download immediately after obtaining URLs
  3. Look for rendition URLs with updated signature parameters

Performance Tips

  • HLS downloads typically run at 10-15x realtime speed
  • No re-encoding with -c copy (fastest)
  • Monitor download with real-time progress display
  • Use absolute output paths to avoid directory confusion

Further Reading

  • PO Token Setup: See references/po-token-setup.md for detailed installation and troubleshooting
  • Internal SOP: references/internal-sop.md — the short list of yt-dlp failures that look like something else (URLs unquoted in zsh, proxy not reaching the PO Token provider). Read it first when a download fails for no obvious reason
  • yt-dlp Documentation: https://github.com/yt-dlp/yt-dlp
  • Format Selection Guide: https://github.com/yt-dlp/yt-dlp#format-selection

Next Step: Transcribe Downloaded Audio/Video

After downloading, if the user's goal involves getting text from the video (transcription, subtitles, meeting notes), proactively suggest:

Download complete: [filename]

If you need the spoken content as text, I can transcribe it for you.

Options:
A) Transcribe with /daymade-audio:asr-transcribe-to-text (Recommended for speech-to-text)
B) No thanks — I just needed the video file

Individual skills in this repo

This repo contains 9 individual skills — each has its own dedicated page.

daymade/claude-code-skills

>- Generates Chinese/Japanese speech with StepFun's Contextual TTS — default stepaudio-2.5-tts, stepaudio-3-tts for whisper/inline-() prosody. Replaces step-tts-2's voice_label with natural-language instruction. Use for emotional/prosody-controlled synthesis, batch voice lines, migrating from step-tts-2, or cloned voices (2.5/step-tts-2, not v3). Triggers on 阶跃 TTS, 语音合成, 配音. Not for transcription (use stepfun-asr).

daymade/claude-code-skills

This skill should be used when comparing two videos to analyze compression results or quality differences. Generates interactive HTML reports with quality metrics (PSNR, SSIM) and frame-by-frame visual comparisons. Triggers when users mention "compare videos", "video quality", "compression analysis", "before/after compression", or request quality assessment of compressed videos.

daymade/local-conversation-history

Entry point for local AI conversation history across providers. Routes a request to the one skill that owns it, by platform (Claude Code, OpenAI Codex, Kimi CLI) and action (read evidence vs continue interrupted work), and owns the one job none of them own alone: a single inventory spanning all three providers. Use when the provider is unknown or plural ("our history", "what have I been working on", "which chats did I have"), when the user wants Kimi CLI history at all, when it is unclear whether they need evidence or resumption, or when they ask for this skill by name. Vague recall that names no platform ("we discussed this once, when was it?") belongs here rather than to a single-provider reader, because a Claude-only answer to an unscoped question cannot support an absence claim. When the platform and the action are both already clear, load that executor skill directly instead — except Kimi CLI, which has no reader or continuation skill of its own and always routes through here.

daymade/ppt-creator

Create professional slide decks from topics or documents. Generates structured content with data-driven charts, speaker notes, and complete PPTX files. Applies persuasive storytelling principles (Pyramid Principle, assertion-evidence). Supports multiple formats (Marp, PowerPoint). Use for presentations, pitches, slide decks, or keynotes.

daymade/prompt-optimizer

Transform vague prompts into precise, well-structured specifications using EARS (Easy Approach to Requirements Syntax) methodology. This skill should be used when users provide loose requirements, ambiguous feature descriptions, or need to enhance prompts for AI-generated code, products, or documents. Triggers include requests to "optimize my prompt", "improve this requirement", "make this more specific", or when raw requirements lack detail and structure.

daymade/qa-expert

This skill should be used when establishing comprehensive QA testing processes for any software project. Use when creating test strategies, writing test cases following Google Testing Standards, executing test plans, tracking bugs with P0-P4 classification, calculating quality metrics, or generating progress reports. Includes autonomous execution capability via master prompts and complete documentation templates for third-party QA team handoffs. Implements OWASP security testing and achieves 90% coverage targets.

daymade/read-claude-code-history

Reads, searches, and exports local Claude Code history without resuming work. Covers recent session inventory, exact session timelines, verbatim human input including queued mid-turn prompts, full-event keyword search, hybrid recall when wording changed, end-state triage, and deleted-file recovery across active Claude homes plus registered archives. Use whenever the user asks what they or Claude said, wants a Claude Code session ID or original context, remembers prior work vaguely, needs an old file from a transcript, or must prove what a Claude session contained before continuing it. Also owns the only Kimi CLI surface, via its Kimi inventory and search flags. For Codex history use read-codex-history; when the request names no platform at all or spans providers, start at local-conversation-history.

daymade/read-codex-history

Reads, searches, and exports local OpenAI Codex history without continuing the old task. Lists recent Codex sessions, extracts exact prompt-ledger inputs by Session, locates a rollout by verified session_meta identity, reconstructs one chronological user/assistant timeline with fork and compaction lineage, and performs bounded keyword search across live and archived rollouts. Use whenever the user asks what they told Codex, wants recent original inputs, a Codex Session ID, full prior context, fork ancestry, or evidence of what a Codex run did. For Claude Code history use read-claude-code-history.

daymade/twitter-reader

Fetch Twitter/X post content including long-form Articles with full images and metadata. Use when Claude needs to retrieve tweet/article content, author info, engagement metrics, and embedded media. Supports individual posts and X Articles (long-form content). Automatically downloads all images to local attachments folder and generates complete Markdown with proper image references. Preferred over Jina for X Articles with images.

Related Skills