Agent skill

Thumbnail Extraction

Extracts the most interesting frames from video files for thumbnail compositing.

Install Thumbnail Extraction

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/thumbnail-extraction .claude/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .claude/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

$skill-installer

$skill-installer install https://github.com/swyxio/skills/tree/main/thumbnail-extraction

Type this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/thumbnail-extraction .agents/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.agents/skills/ instead of .agents/skills for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .agents/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/thumbnail-extraction .cursor/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.cursor/skills/ instead of .cursor/skills for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .cursor/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

gemini skills install

$ gemini skills install https://github.com/swyxio/skills.git --path thumbnail-extraction

--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/thumbnail-extraction .gemini/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .gemini/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction

Installs for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/thumbnail-extraction .github/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .github/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

skills CLI

$ npx skills add swyxio/skills --skill thumbnail-extraction -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install swyxio/skills thumbnail-extraction --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

Manual copy

$ git clone --depth 1 https://github.com/swyxio/skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/thumbnail-extraction .opencode/skills/thumbnail-extraction && rm -rf skills-src

Use ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "thumbnail-extraction" agent skill from https://github.com/swyxio/skills/tree/main/thumbnail-extraction into .opencode/skills/thumbnail-extraction/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "thumbnail-extraction", then confirm the skill loads.

OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.

Facts

Skill name: thumbnail-extraction
GitHub stars: 175
Token cost: ~2.4k tokens
SKILL.md length: 917 words
Files: 3
Skills in repo: 89
Repo updated: Oct 5, 2026
First seen: Oct 7, 2026
Licence: MIT

At a glance

Extracts the most interesting frames from video files for thumbnail compositing.

Works in 3 steps: Quadrant selection (Pass 1 → Pass 2):… → Segment-forced selection (Pass 2 →… → Fallback: If any segment is empty, fill…
Asked to extract thumbnails
SKILL.md covers Overview, When to Use, Dependencies and Pipeline Architecture, plus 5 more sections
Runs Python scripts from its folder; calls python3, pip and pip3; reaches youtube.com

What it does

Thumbnail Extraction is an agent skill from swyxio/skills. Extracts the most interesting frames from video files for thumbnail compositing. Detects faces, expressions, smiles, and presentation slides. Outputs full frames, face crops, and transparent cutouts. Use when asked to extract thumbnails, find interesting frames, grab screenshots from video, or create thumbnail candidates from recordings.

Its SKILL.md is about 2.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `README.md` and `thumbnail_extractor.py`).

It sits in Media & Creative, covering Drug discovery and cheminformatics and Slides and decks. It works with YouTube. The repository describes itself as: Agent skills for Claude Code and other AI agents. The licence is MIT.

When your agent uses it

Asked to extract thumbnails
Find interesting frames
Grab screenshots from video
Create thumbnail candidates from recordings

Example prompts

“Use the thumbnail-extraction skill to extract the most interesting frames from video files for thumbnail compositing”
“/thumbnail-extraction”

Requirements

Python 3

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

Quadrant selection (Pass 1 → Pass 2): Divide video duration into N segments, pick the highest-scoring frame from each segment
Segment-forced selection (Pass 2 → Final): Divide top candidates into top_n equal time segments, pick best from each
Fallback: If any segment is empty, fill from overall top scores

What it can do on your machine

Read from SKILL.md and the folder at commit 038ef34. It shows what the files ask for, not the result of running them.

Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships script files (Python), which the agent can run.
Shell commands in SKILL.md call:
- python3
- pip
- pip3
- yt-dlp
From the folder's file list and the shell code blocks in SKILL.md.
Network
Hosts in commands or code, which the agent is likely to contact:
- youtube.com
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Thumbnail Extraction loads about 2.4k tokens when it runs. Until then it costs about 90 tokens; SKILL.md has 917 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~90
When it runs · the whole SKILL.md, loaded when a task matches: ~2.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from swyxio/skills at commit 038ef34, republished under its MIT licence (© swyxio). 917 words, ~2,392 tokens.

Download SKILL.mdSave it as .claude/skills/thumbnail-extraction/SKILL.md (or your agent's skills folder). This skill also uses 2 other files; get the full folder from GitHub.

name: thumbnail-extraction
description: Extracts the most interesting frames from video files for thumbnail compositing. Detects faces, expressions, smiles, and presentation slides. Outputs full frames, face crops, and transparent cutouts. Use when asked to extract thumbnails, find interesting frames, grab screenshots from video, or create thumbnail candidates from recordings.
metadata.version: 0.1.0

Video Thumbnail Extraction

Overview

Automatically scan a local MP4 video (or YouTube URL via yt-dlp) and extract the 4 most visually interesting frames — prioritizing expressive faces (laughing, shocked, smiling) and engaging presentation slides. Outputs full frames, face crops, and background-removed transparent PNGs ready for compositing.

When to Use

Before creating YouTube thumbnails (feeds into youtube-thumbnails skill)
When you need the best screenshot from a long video recording
When compositing a thumbnail and need transparent guest cutouts
Processing Zoom gallery recordings, interviews, or presentations

Dependencies

Python packages (install once)

bash

# In sandbox (Cowork VM):
pip install opencv-python scenedetect deepface pillow numpy --break-system-packages

# On host Mac (for background removal — sandbox can't download the model):
pip3 install 'rembg[cpu]' pillow --break-system-packages

System tools

ffmpeg (usually pre-installed)
python3 (3.10+)
yt-dlp (optional, for YouTube URLs): pip install yt-dlp --break-system-packages

Model downloads (first run)

DeepFace expression model (~1MB): Downloads automatically on first use. If blocked by proxy, expression detection falls back to OpenCV smile cascade (still effective).
rembg u2net model (~176MB): Downloads on first use. Must run on host Mac if sandbox blocks GitHub releases.

Pipeline Architecture

Two-Pass Design (memory-efficient)

Pass 1 — Quick Scan (OpenCV only, no deep learning)

Sample a frame every 10 seconds across the video
Skip first/last 60 seconds (intro/outro)
For each frame:
- Detect faces via Haar cascade (fast, no GPU needed)
- Detect smiles within face regions
- Compute visual variance (proxy for "interesting" content)
- Detect presentation slides (high edge density + low color saturation)
Score each frame based on: face count, smile count, smile size, visual variance
Select top 12 diverse candidates using quadrant system: divide video into N time segments, pick best from each → ensures temporal spread

Pass 2 — Deep Analysis (DeepFace, only on top 12 candidates)

Re-read only the selected frames from video
Run DeepFace emotion detection (happy, surprise, fear, sad, angry, disgust, neutral)
Weight emotions by thumbnail value: happy > surprise > fear > angry > sad > neutral
Combine Pass 1 score with expression score
Final selection: divide candidates into N time segments, pick best from each → guarantees spread across the full video

Pass 3 — Output (rembg, only on final 4 frames)

Save full frame as JPG (95% quality)
Crop largest detected face with generous padding (0.5x)
Run background removal on face crop → transparent PNG
Generate manifest JSON with metadata

Scoring Heuristics

Signal	Weight	Notes
Face detected	+2.0 per face (cap 3)	Gallery views score high
Smile detected	+3.0 per smile	Cascade-based, no model needed
Smile size ratio	+5.0 × ratio	Bigger smiles = more expressive
Multi-person shot	+1.0 bonus	2+ faces = engaging
Happy expression	+2.0 bonus (Pass 2)	Best for thumbnails
Surprise expression	+2.0 bonus (Pass 2)	Eye-catching
Fear/angry expression	+1.0 bonus (Pass 2)	"Shocked" reactions
Visual variance	+0.0–1.5	Normalized by frame complexity
Presentation slide	baseline 1.5	Useful for slide screenshots

Temporal Diversity Algorithm

The pipeline enforces temporal spread to avoid clustering picks in one segment:

Quadrant selection (Pass 1 → Pass 2): Divide video duration into N segments, pick the highest-scoring frame from each segment
Segment-forced selection (Pass 2 → Final): Divide top candidates into top_n equal time segments, pick best from each
Fallback: If any segment is empty, fill from overall top scores

This ensures a 76-minute video yields picks from different parts (e.g., 1:00, 2:10, 21:50, 48:50) rather than clustering in the most face-heavy section.

Usage

Command Line

bash

python3 thumbnail_extractor.py <video_path> [output_dir] [top_n]

Arguments:

video_path — Path to MP4 file (required)
output_dir — Where to save outputs (default: ~/Downloads/thumb_candidates)
top_n — Number of candidates to extract (default: 4)

Examples:

bash

# Basic — extract 4 best frames
python3 thumbnail_extractor.py "GMT20260130-210038_Recording_gallery_2380x1544.mp4"

# Custom output dir and count
python3 thumbnail_extractor.py recording.mp4 ./thumbs 6

# YouTube video (download first)
yt-dlp -o "video.mp4" "https://youtube.com/watch?v=..."
python3 thumbnail_extractor.py video.mp4

Show full SKILL.md (378 more words)Show less

Output Files

For each candidate, the pipeline generates:

File	Format	Description
`{name}_{n}_{emotion}_{timestamp}_full.jpg`	JPG 95%	Full video frame
`{name}_{n}_{emotion}_{timestamp}_face.jpg`	JPG 95%	Cropped face with padding
`{name}_{n}_{emotion}_{timestamp}_transparent.png`	PNG w/ alpha	Background-removed face cutout
`{name}_manifest.json`	JSON	Metadata for all candidates

Naming example: GMT20260130-210038_3_happy_2-10_full.jpg

GMT20260130-210038 — video name (truncated for Zoom recordings)
3 — candidate number (ranked by score)
happy — detected dominant emotion
2-10 — timestamp (2 minutes 10 seconds)
full / face / transparent — file type

Manifest JSON Structure

json

{
  "video": "GMT20260130-210038",
  "candidates": [
    {
      "index": 1,
      "timestamp": "2:10",
      "timestamp_sec": 130.0,
      "emotion": "happy",
      "emotion_score": 0.85,
      "combined_score": 12.4,
      "num_faces": 3,
      "is_presentation": false,
      "files": {
        "full": "..._full.jpg",
        "face_crop": "..._face.jpg",
        "transparent": "..._transparent.png"
      }
    }
  ]
}

Background Removal (Separate Step)

Since the Cowork sandbox may block model downloads, run rembg on the host Mac:

bash

# On host Mac (via osascript or Terminal)
cd ~/Downloads/thumb_candidates
python3 -c "
from rembg import remove
from PIL import Image
import glob, os

for f in sorted(glob.glob('*_face.jpg')):
    out = f.replace('_face.jpg', '_transparent.png')
    print(f'Processing {f}...')
    img = Image.open(f)
    result = remove(img)
    result.save(out)
    print(f'  -> {out} ({os.path.getsize(out)//1024}KB)')
"

This takes ~10-15 seconds per image on Apple Silicon. The u2net model downloads automatically on first run (~176MB).

Integration with Other Skills

Feeding into `youtube-thumbnails`

After extraction, use the transparent PNGs as compositing elements:

Pick the best face cutout from the candidates
Use it in the Gemini thumbnail prompt as a reference, or
Composite it manually onto the generated Gemini background using ImageMagick:

bash

# Composite transparent face onto Gemini-generated background
convert gemini_background.jpg transparent_face.png \
  -gravity southeast -geometry +50+50 \
  -composite final_thumbnail.jpg

Pipeline flow

[Video MP4] → thumbnail-extraction → [face crops + transparent PNGs]
                                          ↓
                                   youtube-thumbnails → [Gemini background]
                                          ↓
                                   [Composite final thumbnail]

Tuning Parameters

Edit these at the top of thumbnail_extractor.py:

Parameter	Default	Effect
`SAMPLE_INTERVAL_SEC`	10	Lower = more frames scanned, slower
`ANALYSIS_SCALE`	0.5	Lower = faster face detection, less accurate
`SCENE_THRESHOLD`	27.0	Lower = more scene boundaries detected
`MIN_FACE_CONFIDENCE`	0.80	Higher = fewer false positive faces
`top_n`	4	Number of final candidates

For short videos (<10 min), consider SAMPLE_INTERVAL_SEC=5 for finer coverage.

Troubleshooting

OOM / killed process: The v2 pipeline never holds more than 1 frame in memory during Pass 1. If still OOM, increase SAMPLE_INTERVAL_SEC to 15-20.
All emotions "neutral": DeepFace model couldn't download (proxy block). Pass 1 smile detection still works — look at the num_smiles field in the manifest.
Face crop is wrong person: The pipeline picks the largest detected face. In screenshare mode, this may be a profile picture rather than a webcam face. Check the full frame to verify.
No faces detected: Zoom gallery recordings with "shared screen with gallery view" work best. Solo speaker view may have the face too close/large for the cascade detector — try lowering ANALYSIS_SCALE to 0.3.
Background removal artifacts: rembg's u2net can produce halos around hair. For cleaner results, try the u2net_human_seg model: remove(img, model_name='u2net_human_seg').
Slow processing: A 76-minute video takes ~2 minutes for Pass 1, ~15 seconds for Pass 2 (12 candidates), and ~60 seconds for bg removal (4 faces). Most time is in Pass 1 scanning.

Files

SKILL.md and 2 other files in thumbnail-extraction of swyxio/skills.

SKILL.md
README.md
thumbnail_extractor.py

Open the folder on GitHubat commit 038ef34

Compare with similar skills

Thumbnail Extraction next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Thumbnail Extraction compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
Thumbnail Extraction this skillswyxio/skills	175	—	~2.4k	Automated safety check: Pass	MIT	4 days ago
Epub2podcast Gpt Imagedracohu2025-cloud/draco-skills-collection	227	—	~1.9k	Automated safety check: Pass	MIT	22 days ago
Threads Carouselitchernetski/threads-carousel-claude-skill	108	—	~3.5k	Automated safety check: Pass	MIT	5 mo ago
Youtube Notetakersickn33/agentic-awesome-skills	47k	1 repos	~2.3k	Automated safety check: Pass	MIT	today
Hook Writersocial-media-skills/skills	128	—	~1.7k	Automated safety check: Pass	MIT	8 days ago
Nlm Skilliusztinpaul/ai-research-os-workshop	179	1 repos	~6.9k	Automated safety check: Pass	MIT	3 mo ago

Similar skills

Epub2podcast Gpt Image
dracohu2025-cloud/draco-skills-collection
可独立运行的 GPT-Image 增强版 EPUB2Podcast：在本地把 EPUB 转成双人中文音频、GPT-Image/Smart Slide 视觉页、最终 MP4，并生成 YouTube 发布素材。
227 GitHub stars~1.9k tokensUpdated 22 days ago
Media & CreativeAuto-check passed
Threads Carousel
itchernetski/threads-carousel-claude-skill
Convert text posts into visual carousel images or presentations for Threads, Instagram, LinkedIn, TikTok, YouTube.
108 GitHub stars~3.5k tokensUpdated 5 mo ago
Documents & OfficeAuto-check passed
Youtube Notetaker
sickn33/agentic-awesome-skills
Turn YouTube talks into local study notes with slides, transcripts, editable annotations, and a markdown-backed viewer.
47k GitHub starsUsed in 1 repo~2.3k tokens
Documents & OfficeAuto-check passed
Hook Writer
social-media-skills/skills
A skill your agent uses to write the hook — the opening that earns attention — for any social content: a caption's first line, a video's first three seconds, a carousel cover slide, a thread opener…
128 GitHub stars~1.7k tokensUpdated 8 days ago
Media & CreativeAuto-check passed
Nlm Skill
iusztinpaul/ai-research-os-workshop
Expert guide for the NotebookLM CLI (nlm) and MCP server - interfaces for Google NotebookLM.
179 GitHub starsUsed in 1 repo~6.9k tokens
Knowledge ManagementAuto-check passed
Brand and Design Toolkit
nextlevelbuilder/ui-ux-pro-max-skill
Bundles design tasks behind one skill: brand identity, tokens, UI styling, logos, corporate identity mockups, slides, banners, icons and social images.
134k GitHub starsUsed in 1 repo~3.5k tokens
Media & CreativeAuto-check passed

Questions about Thumbnail Extraction

What does Thumbnail Extraction do?

Extracts the most interesting frames from video files for thumbnail compositing. Thumbnail Extraction is an agent skill from swyxio/skills. Extracts the most interesting frames from video files for thumbnail compositing.

When should I use Thumbnail Extraction?

Thumbnail Extraction fits situations like: asked to extract thumbnails; find interesting frames; grab screenshots from video; create thumbnail candidates from recordings.

How do I install Thumbnail Extraction in Claude Code?

Run `npx skills add swyxio/skills --skill thumbnail-extraction -a claude-code`. Or copy the skill folder (thumbnail-extraction in swyxio/skills) into .claude/skills/thumbnail-extraction in your project. Claude Code loads it when a task matches its description.

How do I install Thumbnail Extraction in Codex?

Run `npx skills add swyxio/skills --skill thumbnail-extraction -a codex`. Or copy the skill folder (thumbnail-extraction in swyxio/skills) into .agents/skills/thumbnail-extraction in your project. Codex loads it when a task matches its description.

Can I use Thumbnail Extraction in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add swyxio/skills --skill thumbnail-extraction -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/thumbnail-extraction, .gemini/skills/thumbnail-extraction, .github/skills/thumbnail-extraction and .opencode/skills/thumbnail-extraction in your project.

What does Thumbnail Extraction need to run?

Going by SKILL.md and its folder, Thumbnail Extraction needs Python for the scripts in its folder and the command-line tools its instructions call (python3, pip, pip3 and yt-dlp). Our summary lists: Python 3.

Does Thumbnail Extraction access the network?

SKILL.md names 1 domain. In commands or code: youtube.com; the agent is likely to contact it when it follows the instructions. This is read from the text; nothing was executed.

Is Thumbnail Extraction safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Thumbnail Extraction use?

Thumbnail Extraction is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Thumbnail Extraction use?

About 2.4k tokens (SKILL.md is roughly 9.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Thumbnail Extraction?

Skills that share tags, products or a category with Thumbnail Extraction: Epub2podcast Gpt Image (dracohu2025-cloud/draco-skills-collection, 227 stars), Threads Carousel (itchernetski/threads-carousel-claude-skill, 108 stars), Youtube Notetaker (sickn33/agentic-awesome-skills, 47k stars) and Hook Writer (social-media-skills/skills, 128 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Thumbnail Extraction?

swyxio (a GitHub user) maintains it in swyxio/skills, which has 175 GitHub stars. The repository holds 89 skills in this directory. The repository was last updated on October 5, 2026.

Source: swyxio/skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Install Thumbnail Extraction

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Video Thumbnail Extraction

Overview

When to Use

Dependencies

Python packages (install once)

System tools

Model downloads (first run)

Pipeline Architecture

Two-Pass Design (memory-efficient)

Scoring Heuristics

Temporal Diversity Algorithm

Usage

Command Line

Output Files

Manifest JSON Structure

Background Removal (Separate Step)

Integration with Other Skills

Feeding into youtube-thumbnails

Pipeline flow

Tuning Parameters

Troubleshooting

Files

Compare with similar skills

Similar skills

More from swyxio/skills

Related

Questions about Thumbnail Extraction

Feeding into `youtube-thumbnails`