Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Minimax H3 Storyboard Director

    2

    Transforms scripts into cinematic storyboards and MiniMax H3 prompts with strict continuity and camera logic.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Minimax H3 Storyboard Director

    Minimax H3 Storyboard Director

    Example session with this skill installed

    Convert this 30-second dialogue script between a detective (S1) and a witness (S2) in a rainy alley into a MiniMax H3 storyboard and prompt package. S1 is wearing a wet trench coat.

    • Read your context and instructions
    • Compiled the minimax h3 storyboard

    The script has been processed into a cinematic storyboard and H3 prompt package. All dialogue is preserved, and a continuity ledger has been established for the rainy environment and S1's wet trench coat.

    Cinematic Storyboard

    ShotFraming/LensMovementVisual DescriptionAudio/Dialogue
    01Wide Shot, 35mmSlow Push-inDeep depth of field. A narrow, dark alley glistening with rain. S1 stands back to camera, wet trench coat reflecting neon lights.Heavy rain hitting metal bins.
    02Medium Close-upHandheldShallow depth. S1 turns slightly, rain dripping from his hat. S2 is visible in the background, out of focus.(S1): "You weren't supposed to be here."
    03Close-upStaticFocus on S2's trembling hands holding a damp cigarette. High contrast lighting.(S2): "I had nowhere else to go."

    MiniMax H3 Prompt Package

    [subject_definitions]
<Subject 1> is a detective in a wet, dark brown trench coat.
<Subject 2> is a witness in a grey hoodie, looking anxious.

[summary]
A cinematic dialogue scene in a rainy alley at night with high-contrast neon lighting and realistic water physics.

[retention_analysis]
Maintain S1's wet coat texture and S2's hand tremors across shots. Rain intensity remains constant.

[detailed_description]
The scene opens with a wide shot of a dark, narrow alleyway. The ground is soaked, reflecting flickering blue neon signs. Hard cut to a medium close-up of (S1), his wet trench coat showing detailed water droplets. He turns his head slowly. (S1): <d>[English] You weren't supposed to be here.</d> The camera shifts to a tight close-up of (S2)'s hands. The hands are shaking while holding a cigarette. (S2): <d>[English] I had nowhere else to go.</d>

[overall_soundscape]
Continuous sound of heavy rain falling on asphalt and metallic surfaces. Low-frequency city hum in the background.

[non_diegetic_music]
None.

    Next steps

    1. Map <Subject 1> to your uploaded character reference image.
    2. Confirm if you want to use a specific voice ID for the detective's dialogue.
    3. Review the "Detailed Description" for any specific prop additions like the cigarette.

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

    About this skill

    Production scripts often fall apart during the AI video generation phase due to inconsistent character continuity, generic camera work, or broken prompt syntax. This skill bridges the gap between creative screenwriting and the technical MiniMax H3 Ref2VA workflow. It functions as a virtual Director of Photography, converting scripts into structured cinematic storyboards and valid H3 prompt packages.

    What it does

    • Cinematic Storyboarding generates detailed Chinese-language shot lists focusing on physical camera logic, blocking, and depth of field without unnecessary timestamps.
    • H3 Prompt Engineering writes six-section English prompts optimized for MiniMax H3, including subject definitions, soundscapes, and retention analysis.
    • Continuity Tracking maintains a state ledger for characters, props, clothing, and hand occupancy across multiple shots to prevent AI "hallucinations."
    • Camera Logic Enforcement applies specific lens focal lengths, depth-of-field strategies, and movement types to ensure visual variety and narrative impact.
    • Reference Management maps subjects to specific picture, video, and audio reference IDs to lock in character identity and voice.

    How it works

    1. Fact Locking identifies characters, props, and exact dialogue from the source script.
    2. Spatial Pre-viz defines the environment dimensions, lighting sources, and character action paths.
    3. Storyboard Design builds the visual sequence using cinematic terminology and emotion-driven cutting points.
    4. H3 Encoding generates the final prompt package using the detailed_description syntax and reference material declarations.

    Frameworks & tools

    This skill is designed for the MiniMax H3 video generation model, specifically utilizing the Ref2VA (Reference to Video/Audio) workflow and h3DigitalHuman modules.

    Why this beats prompting it yourself

    Generic prompts often result in "talking heads" or characters that change clothes between shots. This skill implements a strict state ledger and 13 core cinematic rules, such as limiting static shots to 30% and enforcing physical depth-of-field strategies that standard LLMs ignore.

    Use cases

    • Converting a narrative script into a shot-by-shot production plan for AI video.
    • Creating music video storyboards that synchronize lyrics with specific character lip-sync prompts.
    • Reviewing existing H3 prompt packages for continuity errors or broken camera logic.

    Known limitations

    Does not support automatic timestamp generation by default. Requires manual reference asset IDs (e.g., Picture 1, Subject 2) to be provided or pre-defined.

    Does not generate timestamps by default. Requires manual mapping of reference asset IDs. Optimized for live-action realism rather than stylized animation.

    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 19 days ago

    Software
    Git

    Permissions

    Write Files
    Listed19 days ago

    What's inside

    Frequently Asked Questions

    This skill acts as a virtual Director of Photography, converting narrative scripts into structured shot lists and technical H3 prompts that maintain consistent character appearance and camera logic across multiple clips.

    The skill generates a 'state ledger' that tracks character clothing, prop placement, and hand occupancy, ensuring that a character's look remains identical from the first shot to the last.

    It is specifically optimized for the MiniMax H3 model and its Ref2VA (Reference to Video/Audio) workflow, utilizing the detailed_description syntax and h3DigitalHuman modules.

    Upon purchase, you receive the full agent configuration, a library of 13 core cinematic rules for prompt engineering, and the character state ledger template.

    Yes, while it provides the prompt structure, you will need to map your own image or audio IDs (e.g., Picture 1, Audio A) into the provided placeholders to link the AI to your specific assets.

    The skill focuses on physical blocking and camera logic; while it supports narrative pacing, it does not currently generate automated timestamps for the video segments.

    New in Image & Video