Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Content Spy

    1

    Use it only for your own videos or ones you have permission to analyze.

    $10

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Content Spy

    Content Spy

    Example session with this skill installed

    /contentspy single https://www.youtube.com/shorts/xxxxxxxxxxx make it work for my tech channel

    • Read your context and instructions
    • Compiled the content spy

    Example output for an illustrative video. Your own results come from real data.

    Summary

    A 38-second "warning" review of a budget phone built as three quick tests with a verdict at the end. It works because the hook pairs a warning with a price in under 2 seconds, and every claim is shown within about a second.

    Hook (0–3 s)

    First words"Don't buy this $300 phone until you see what happens to the battery."
    First framePhone held up to the camera, screen at 100 % battery
    On-screen text"DON'T BUY THIS ($300)"
    Hook typeWarning + price shock + curiosity gap
    Cuts in first 5 s4

    Pacing

    MetricValue
    Words per minute182
    Cuts per minute34.7
    Average shot1.65 s
    Dead airone 1.6 s pause before the reveal (deliberate)

    Beat-by-beat

    TimeWhat's saidWhat's shownPurpose
    00:00"Don't buy this $300 phone…"Phone to camera + red bannerHook
    00:07"Test one: camera…"Side-by-side photosProof
    00:26(pause) "Nine percent."Battery screenPayoff
    00:34Verdict that loops into the first lineFace to cameraLoop ending

    Reusable script

    [0–2s] HOOK: "Don't buy this [PRICE] [PRODUCT] until you see what happens to the [WEAK SPOT]."
    [2–6s] SETUP: "On paper it's perfect: [SPEC 1], [SPEC 2]."
    [6–25s] TESTS: claim → proof shot → next claim
    [25–28s] PAYOFF: pause, then "[shocking number]."
    [last 2s] LOOP: verdict that flows back into the hook

    Next steps

    • Film the hook with your product in frame from the first frame
    • Keep a proof shot within 1 second of every claim
    • Open report.md for the full breakdown and adaptation ideas

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Find out why a video works, then make your own

    Ever watched a video blow up and wondered why? ContentSpy breaks down any video, short or long, and turns what made it work into a script you can reuse.

    Works with: YouTube (videos and Shorts), TikTok, Instagram, X and Threads, or a video file on your computer.

    Two commands

    /contentspy single <video link>: a full breakdown of one video

    • Hook: the first 3 seconds, frame by frame. Exact first words, first frame, on-screen text, and why it stops the scroll
    • Beat-by-beat structure with timestamps (what's said, what's shown, and why)
    • Pacing: words per minute, cuts per minute, shot lengths, dead air
    • Narration and voice: tone, catchphrases, openers and sign-offs
    • Sound and look: loudness, music bed, colour palette, brightness and motion
    • Retention tricks, call to action, title and thumbnail
    • A fill-in-the-blank script template and ideas adapted to your channel

    /contentspy channel <creator>: a creator playbook Studies the creator's 10 most popular and 10 latest videos, then shows their hook formulas side by side, their structure template, narration style, editing recipe, title formulas, posting rhythm, and what changed recently.

    What you get

    A detailed report.md (or playbook.md) saved in a reports folder, with contact sheets of the video and a facts.md holding every measured number, plus a short summary in chat.

    Built for every creator

    • Speech-to-text runs locally, so it also works for videos without captions and for languages most tools ignore (such as Amharic)
    • Runs on your own computer; your files stay local
    • Open source (MIT): github.com/Abel9436/ContentSpy

    Note: fetching a video to analyze its visuals is opt-in. Use it only for your own videos or ones you have permission to analyze.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Social Media.

    Requires ffmpeg for visual analysis. Instagram/X often need browser cookies. Visual analysis requires local files or explicit download opt-in. Public tags for Threads provide limited metadata.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 11 days ago

    Accounts
    Hugging Face
    Software
    Python 3.10+FFmpegyt-dlpgallery-dlGit
    Optional
    faster-whisper

    faster-whisper adds speech-to-text.

    Permissions

    Terminal / Shell
    Write Files
    Read Files
    Network Access
    Environment Variables

    Allowed Hosts

    huggingface.co

    File Scopes

    ~/ContentSpy/reports/**
    (where
    reports
    are
    saved)
    ~/.claude/skills/contentspy/**
    (the
    skill
    and
    its
    settings
    ~/.claude/settings.json
    (setup
    adds
    permission
    rules
    for
    the
    skill’s
    own
    scripts)
    Listed11 days ago

    What's inside

    Frequently Asked Questions

    ContentSpy provides objective metrics like cuts per minute, words per minute, and frame-by-beat logs, whereas standard LLMs can only analyze the text transcript without understanding the visual pacing or production style.

    The skill supports a wide range of platforms including YouTube, TikTok, Instagram Reels, X (Twitter), and Threads, as well as local video files for pre-release auditing.

    Yes, ContentSpy requires 'ffmpeg' for scene detection and 'yt-dlp' for metadata extraction; ensure these are installed on your host system for the skill to process visual data correctly.

    The 'Channel' command automatically scans a creator's 10 most popular and 10 most recent videos to identify the specific 'DNA' and recurring patterns that drive their audience engagement.

    For platforms with strict anti-scraping measures like Instagram or X, you may need to provide browser cookies to allow the skill to authenticate and retrieve video metadata successfully.

    New in Social Media