More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Shotpack: AI video shots that keep the same look

    1

    A timestamped shot list, one prompt block per shot with look and motion split, stills approved before video credits, takes one job at a time, two or three music mixes, frame QA and per-ratio exports.

    $15

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Shotpack: AI video shots that keep the same look

    Shotpack: AI video shots that keep the same look

    Example session with this skill installed

    Shotpack: 20-second AI news-anchor teaser for our product launch, in 9:16 and 16:9. We use Higgsfield. The presenter reference photo is refs/anchor.jpg.

    • Read your context and instructions
    • Compiled the shotpack

    Access: Higgsfield allowed, 240 credits (example balance), 1 approved reference photo.

    Shot list (5 shots, 20 s)

    #TimeA/BJobFramingCameraDuration
    10:00-0:03ACredibility openMedium close-upFixed3 s
    20:03-0:07AHeadlineMedium close-upFixed4 s
    30:07-0:12BSupport: product on a deskClose-upSlow push5 s
    40:12-0:16AKey lineMediumFixed4 s
    50:16-0:20BPayoff closeWideFixed4 s

    3 A-roll, 2 B-roll. A-roll uses image-to-video from approved stills.

    Continuity anchors: navy blazer, short dark hair, warm key light, slate-blue studio backdrop.

    SHOT 2 | 0:03-0:07 | A-roll | job: headline | start image: still-02 | 9:16 | 4s
    Look: presenter, navy blazer, short dark hair, medium close-up, slate-blue studio, warm key light.
    Motion: natural blink, subtle head turn to camera, steady eye contact, camera steady.
    Avoid: on-screen text, logos, extra people, face or outfit change.
    Overlay (edit): "Launching today". Voice: "Something new arrives today." Cut on the beat.
    

    Stills: 5 generated; still-04 failed hand QA and was regenerated. No video credits spent yet.

    Next steps

    • Approve stills 01 to 05.
    • Pick one of the 3 music genres I propose for this audience.
    • I then run 2 to 3 takes per shot, one job at a time.

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

    About this skill

    AI video generators give you a different face in every shot, garbled lettering inside the frame, and a credit bill for looks nobody approved. One overloaded prompt tries to fix the look and the motion at once. Shotpack plans, generates and checks short AI videos so the shots keep the same look and video credits go only to approved stills.

    What it does

    • Access check first: allowed generators, credit balance and approved reference photos, or it stops.
    • Timestamped shot list in A-roll and B-roll, one job per shot, with text and logos kept as edit overlays.
    • Prompt pack per shot with the look and the motion written separately, continuity anchors repeated in every block, an avoid list and a copy-paste line for the tool UI.
    • Stills before video credits: start frames are QA'd against the reference and approved by you, with IDs recorded in continuity.md.
    • Takes one job at a time, 2 to 3 per shot, each logged in prompts-log.md.
    • Mixes, QA and exports: 2 to 3 music mixes at about -14 LUFS with LICENSES.md, frame checks, and exports in 9:16 plus 16:9 or 4:5.

    How it works

    1. Lock the objective in discuss mode; "strategy first" stops before any prompt.
    2. Shot list and prompt pack, each shot set to image-to-video or text-to-video on purpose.
    3. Stills, then takes: approval gate before video, contact sheet of sample frames to pick from.
    4. Assemble and score: silent master, cuts.txt, music and SFX at hit points.
    5. QA and export: faces, hands, in-frame text, ffprobe, black and freeze checks, OCR, then reframed exports with a cover.

    Frameworks & tools

    A video generator the owner allows (for example Higgsfield), an image model for stills, and ffmpeg for frames, assembly and encoding (libx264, yuv420p, +faststart). Music from licensed tracks or code (numpy/scipy, FluidSynth).

    Why this beats prompting it yourself

    Typing one prompt per clip spends credits on guesses. Shotpack splits look from motion so a failure is fixed in one part, holds video credits until stills are approved, and logs every take so a good result can be repeated.

    Use cases

    • Ads and teasers from 3 to 5 short shots.
    • AI news-anchor segments.
    • Contest entries of 30 to 45 seconds.
    • Animated stills with text as overlays.

    Known limitations

    Needs a generator you allow and credits to spend. Generator connector audio is speech-only, so music and SFX come from licensed tracks or code. Not for films built from real app UI, transcription or editing real footage.

    Needs a generator you allow and credits to spend. Generator connector audio is speech-only, so music and SFX come from licensed tracks or code. Not for films built from real app UI, transcription or editing real footage.

    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 any payment.

    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 29 days ago

    Requires

    Higgsfield
    ffmpeg
    ffprobe
    numpy/scipy
    FluidSynth

    Permissions

    Write Files
    Listed29 days ago

    What's inside

    Frequently Asked Questions

    Access to a video generator you allow, enough credits, and approved reference photos. It checks all three first and stops if any is missing.

    Video credits are spent only after you approve the stills, takes run one job at a time, and it does not re-render without your direction.

    It plans a timestamped shot list, writes look and motion separately per shot, repeats continuity anchors so faces hold, logs every take, and checks every frame before export.

    The zip contains SKILL.md (the instructions your agent follows) and LICENSE.md. Nothing else is installed. No generator access or credits are included.

    New in Image & Video