More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Take a Video Apart, Shot by Shot

    1

    Take a reference video apart shot by shot — what each shot is, how it was made, what the audio is doing — and turn it into a plan for making your own version.

    $12

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Take a Video Apart, Shot by Shot

    Example session with this skill installed

    Analyze the 15-second mobile app promo video I've provided. I need a full shot log and a remake plan where we replace the filmed hand-held shots with AI-generated product lifestyle shots.

    • Read your context and instructions
    • Compiled the take a video

    I have deconstructed the 15-second promo video into 4 distinct shots. I have classified each shot based on visual evidence and mapped out a remake plan that replaces the original camera footage with generative assets while maintaining the coded UI overlays.

    Video Teardown & Remake Plan

    Scope Note: This report is an analysis for learning and creating an original remake. All original footage and audio remain the property of the creator; this plan is for an original version using new assets.

    Shot Log & Classification

    ShotTimecodeTypeMethodEvidence / Notes
    10:00-0:04MediumFilmedHand-held shake at 0:02; natural light falloff on device screen.
    20:04-0:07Close-upCodedPixel-perfect UI transitions; no motion blur on button hover at 0:05.
    30:07-0:12WideFilmedReal-world shadows on desk surface; depth of field blur at 0:09.
    40:12-0:15Text CardCodedVector-clean typography; precise 60fps easing on entrance.

    Audio Split

    • Voice: Synthetic AI male voiceover, studio-dry, peaks at -3dB.
    • Music: Lo-fi hip hop beat, 90 BPM; hard cuts land on every 4th beat.
    • Effects: Haptic "click" at 0:05 (Shot 2) and "swoosh" transition at 0:07.

    Remake Plan

    Shot 1 & 3 (Generate): Use a generative video model to create "High-end minimalist office, soft morning light, product on desk." Replace filmed hand-held movement with a smooth digital pan.
    2. Shot 2 & 4 (Code): Re-export UI animations from the design file at 60fps to match the original timing.
    3.

    Asset Prep: Record new voiceover and source an original royalty-free track at 90 BPM to maintain the edit's rhythm.

    Next steps

    • Batch generate the lifestyle assets for Shots 1 and 3.
    • Export the UI interaction sequence for Shot 2 using the identified timing.
    • Align the new audio track to the existing 4-beat cut structure.

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

    About this skill

    Most people watch a reference video and come away with a feeling — fast, polished, funny. Feelings do not remake anything. A teardown replaces the feeling with evidence: every shot logged, every claim tied to a frame, every sound accounted for — so the video can be deliberately remade, by you or by an agent, as your own original version. The output is not a summary. It is a working plan: per shot, what to film, what to generate, and what to code, with the proof attached.

    Inside the package: the skill file with the six steps and the scope rules, a teardown report template with the per-shot fields and evidence rules, and a complete fictional teardown with its remake plan. The honesty rule sits at the top of the skill and in the header of every report: a teardown is analysis for learning and for making an original remake. It is not a license to redistribute the original footage, lift its audio, or pass its shots off as yours. Work on material you have the right to study; where ownership of an element is unclear — a music track, a distinctive performance — the plan flags it as replace, never as reuse.

    A run goes like this: segment the video into shots on the evidence — watch at low speed, step frame by frame at every suspected cut, mark exact in and out timecodes, save frames from the start, middle, and end of each shot, and name every transition, because hidden transitions are where careless teardowns merge two shots into one and ruin the plan. Log each shot with the same fields every time: number and timecodes, type in plain words, what is on screen, on-screen text transcribed exactly with its timing, and how the shot arrives and leaves. Classify how each shot was made — filmed (look for natural light falloff, real depth of field, shadows that agree), AI-generated (morphing detail, text that will not hold still, physics that almost works), or coded or screen-captured (pixel-perfect interface, perfectly repeatable motion) — citing frames for every call, with mixed and uncertain as honest, allowed answers. Give the audio its own pass rather than half-hearing it while staring at frames: voice, music, and effects, each tied to shots and moments, including the mix — what sits on top, what ducks under voice, what carries the transitions — because remakes fail on mix more often than on missing sounds. Only then write the plan: per shot, the make method, the concrete setup and text and timing, what must exist before that shot can be made, a rough effort grade, and its audio — closing with a build order and the full list of everything flagged replace or uncertain. A variant covers recreating a single moving element — a website animation is the classic case — frame by frame: measure position, size, opacity, and timing from the frames, rebuild it in code from those measurements, and compare against the original at matched frames until no difference you can name remains.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Requires manual frame-stepping to identify transitions. Output is a text-based production plan, not a rendered video. Does not handle legal licensing for original reference materials.

    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 6 days ago

    Permissions

    No special permissions declared or detected

    Listed6 days ago

    What's inside

    Frequently Asked Questions

    You will receive the proprietary system prompts, a structured Markdown reporting template, and a workflow guide designed for frame-accurate analysis and remake synthesis.

    While this skill works with most large language models, it is optimized for multimodal agents that can process image sequences or frame-by-frame data for precise evidence-based logging.

    It solves the problem of 'vague interpretation' by providing a technical blueprint that identifies exactly how a shot was made (filmed vs. AI-generated) and the specific easing curves needed to replicate the motion.

    No, this skill generates a comprehensive technical report and execution plan; you will still need to use your preferred editing or generation tools to create the actual video assets.

    Yes, the framework is designed to handle everything from live-action product demos to complex motion graphics and AI-generated social content by focusing on frame-level evidence.

    Setup is straightforward: you import the provided Markdown framework and system instructions into your AI agent's environment to begin the structured analysis process.

    New in Agents & Orchestration