More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Ctrl-H Video Kit V3.0.0

    1

    A customizable video production skill that guides Claude through creating professional, voiceover-synced explainer videos using a Python animation engine, with dynamic visuals, sound effects, and 16:9 or 9:16 exports.

    Free

    Security scanned
    Ctrl-H Video Kit V3.0.0

    Ctrl-H Video Kit V3.0.0

    Example session with this skill installed

    Ctrl + H Video Kit V2.1.0

    Repository: https://github.com/heshamadel1/ctrl-h

    Watch what the kit does: https://youtu.be/bwPKh-Q9y2U

    A branded explainer-video pipeline plus Claude skills. Give it a voiceover WAV and a declarative scene file; it renders a synced

    16:9 (1920x1080) or

    9:16 (1080x1920) video at 30 fps with a library of 34 sound effects, optional Arabic captions and an optional thumbnail. It started for the Ctrl + H programming channel and works for any channel: on first use Claude asks for your identity, colours, cards and background and saves them in ctrlh.config.json.

    What you get

    3 themesorange (dark), paper_orange (cream paper, thick borders), soft_cream (white rounded cards, soft shadow, orange glow)
    25 backgroundsanimated hud_live, 9 dark styles, paper, cream, 10 technology backgrounds (tech:laravel...), 3 ready images; picked per video
    12 card styles + 100+ blockscards, code, terminal, flow, architecture, ER, HTTP, diff, queue, polls, chat window, mini charts, counters
    34 sound effectsassets/sfx, placed automatically at 30% under the voice or pinned by name
    Captions (optional)Arabic in a box, orange highlight on key words and every English word
    Two sizes16:9 and 9:16, both native renders of one scene file
    One commandbuild.py: lint, render, SFX, mux, meta.md, chapters.txt; long videos split into parts
    Safetylint.py (layout, timing, logo, features), check_voice.py (WAV vs scene file), smoke.py (whole-kit test), sfx_preview.py (hear the sounds), all without rendering a video
    Clipsreels.py cuts 30-60 s Reels/Shorts from a long video
    Skillsctrlh-video-style (asks before every video) and ctrlh-thumbnail-style

    Quick start

    git clone https://github.com/heshamadel1/ctrl-h.git && cd ctrl-h
    python3 -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt
    cd examples && FMT=wide python3 ../engine/render.py template_howto_scenes frames /tmp/frames 30
    python3 ../engine/build.py my_scenes --vo voice.wav --out out      # a full video
    

    Needs Python 3.13, FFmpeg, Pillow with RAQM. Details: Getting started.

    Documentation

    FileRead it for
    GETTING_STARTED.mdinstall, first frame, first video, Claude skills
    PIPELINE.mdthe one workflow (automated and manual), CLI flags, SFX, captions, formats
    CONFIGURATION.mdevery config key / env variable, priority, themes, backgrounds
    CATALOG.mdevery card, block, theme, background and sound by name and code
    FEATURES.mdevery feature with a picture and its fields
    SOUNDS.mdthe 34 sound effects
    CUSTOMIZE.mdwhat to edit for what; using it for your own channel
    ARCHITECTURE.mdmodules and data flow
    TROUBLESHOOTING.mdcommon errors and fixes
    CONTRIBUTING.mdhow to contribute
    CHANGELOG.md, RELEASE_NOTES.md, RELEASE_STEPS.mdversions and publishing

    With Claude

    Install skills/ctrlh-video-style and skills/ctrlh-thumbnail-style, then send a script and a WAV. Claude asks, every video: theme, size (16:9 / 9:16), captions, cards, background, sound effects (specific names or automatic), thumbnail.

    License

    Code: MIT (LICENSE). Fonts: SIL Open Font License. The Ctrl + H name and logo are the channel owner's brand assets; do not reuse them without permission.

    • Read your context and instructions
    • Compiled the ctrl-h video kit

    Expected Result — Complete Video Production Deliverable

    The agent delivers a complete explainer video production project using the Ctrl + H Video Kit. The deliverables must clearly distinguish between generated artifacts and outputs that have actually been executed, rendered, or verified.

    1. Production Summary

    Create a professional animated explainer video about

    "How Laravel Middleware Works", using the supplied Egyptian Arabic voiceover with English technical terminology.

    Follow the Ctrl + H visual identity:

    • Cinematic black backgrounds.
    • Orange accents (#FF8A00) and white typography.
    • Clean Arabic typography and readable coding fonts.
    • Smooth transitions and meaningful technical visualizations.

    2. Generated Outputs

    These are the artifacts created by the agent. Their generation does not imply successful execution or verification.

    DeliverableExpected Output
    Voiceover AnalysisNarration analysis and proposed timing breakdown
    Scene PlanComplete ordered scene descriptions and visual objectives
    Animation Source CodeComplete Python scene scripts and supporting code
    Visual AssetsGenerated diagrams, backgrounds, cards, and other visual resources
    Audio IntegrationAudio configuration and synchronization instructions
    Sound DesignSelected sound effects and integration instructions
    Project StructureOrganized source files, assets, and configuration
    Build InstructionsExact commands and dependencies required to build the video

    3. Verified Outputs

    Only list an item here after performing the corresponding check and observing its actual result.

    VerificationRequired Evidence
    Source CodeGenerated scripts pass applicable syntax and validation checks
    DependenciesRequired dependencies are installed or their availability is confirmed
    Scene RenderingAll intended scenes render successfully
    Audio SynchronizationRendered scenes align with the supplied voiceover
    Sound EffectsIntegrated effects play correctly and are appropriately balanced
    Video ExportFinal video file exists and can be opened
    Video SpecificationsActual resolution is 1920×1080 and frame rate is 30 FPS
    Visual QualityRendered output has been inspected for missing assets, visual glitches, and incorrect transitions

    If a check cannot be performed, mark it as

    Not Verified. If it fails, report the failure and its details rather than presenting it as successful.

    4. Scene Breakdown

    Provide a complete scene-by-scene production table.

    SceneContentVisual Direction
    01Introduction to Laravel MiddlewareAnimated request entering the application
    02The Request LifecycleRequest pipeline flow diagram
    03Middleware ExecutionMiddleware layers processing a request
    04Request FilteringAllowing or rejecting requests
    05Route ProtectionAuthentication middleware protecting a route
    06Practical ExampleAnimated Laravel middleware code walkthrough
    07SummaryRecap of the key concepts

    Actual scene durations and transitions must follow the supplied voiceover, not arbitrary predefined timings.

    5. Complete Project Delivery

    Provide

    • Complete generated source code, not truncated snippets or pseudocode.
    • All generated assets and supporting resources.
    • Audio synchronization data and configuration.
    • Build scripts and exact execution commands.
    • Final project directory structure.
    • Clear documentation of any missing dependencies or external resources.

    6. Final Status Report

    End the delivery with three explicit categories:

    Generated: List every artifact successfully created.

    Verified: List only artifacts that were actually executed, rendered, inspected, or tested, together with the verification evidence.

    Pending / Failed: List anything not yet verified, unavailable, or unsuccessful, including the reason and the next action required.

    7. Next Steps

    1. Provide the complete project files.
    2. Explain how to execute the build process.
    3. Identify outstanding dependencies or assets.
    4. Provide the final rendered video location, if rendering succeeded.
    5. Recommend improvements based on actual inspection.

    Completion Criteria: A generated artifact must never be described as verified without evidence. A final video must never be described as rendered unless the file was actually produced, and it must never be described as quality-checked unless it was actually inspected.

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

    About this skill

    Ctrl + H Video Kit V3.0.0

    Release Claude Free first Advanced agents optional Formats License: MIT

    Independent open-source project. Not affiliated with or endorsed by Anthropic or OpenAI.

    Repository: https://github.com/heshamadel1/ctrl-h · Watch what the kit does: https://youtu.be/8sf53wNu3yo

    A branded explainer-video pipeline plus Claude skills. Give it a voiceover WAV and a script; it renders a synced Wide 16:9 (1920x1080) or Reel 9:16 (1080x1920) video at 30 fps with a library of 34 sound effects, optional Arabic captions and an optional thumbnail. It started for the Ctrl + H programming channel and works for any channel: on first use Claude asks for your identity, colours, cards and background and saves them in ctrlh.config.json.

    Who it is for: Claude Free users first

    The Kit is designed first of all for regular Claude (Free) users. The Standard Claude Workflow needs only a Claude chat, your script and your voiceover WAV. It does not need Claude Code, a terminal, local development tools, a paid subscription, an external agent, Whisper or any advanced tooling: Claude runs the Kit in its own workspace and hands you the finished video.

    Standard Claude Workflow (primary)Advanced Agent Workflow (optional)
    ForAnyone using Claude in the app or on the webDevelopers using Claude Code, OpenAI Codex or another coding agent
    You doInstall the skill, send a script and a WAV, answer Claude's questionsSame, and the agent can also run commands, tests and optional tools on your machine
    NeedsA Claude chat that can use skills and run code (availability depends on your Claude plan and is set by Anthropic)An agent environment with shell access
    Optional extrasNoneWhisper alignment, layout_check.py, storyboard.py validate, golden tests, perf profiling
    OutputIdenticalIdentical (an optional tool never changes how the video looks)

    Read docs/WORKFLOWS.md for the exact steps and the list of what is standard and what is advanced.

    How to start (Standard Claude Workflow)

    1. Add the skills from skills/ (ctrlh-video-style, and ctrlh-thumbnail-style if you want thumbnails) to your Claude, then attach this Kit (the ZIP) to the chat if Claude asks for it.
    2. Send your script and your voiceover WAV (Arabic, English or mixed).
    3. Answer Claude's questions (it asks every video): theme, size (Wide or Reel), captions, cards, background, sound effects, thumbnail.
    4. Claude builds the video and sends it back. Ask for changes in plain words ("shorter cards", "make the reel version", "different transition").

    What you get

    3 themesorange (dark), paper_orange (cream paper, thick borders), soft_cream (white rounded cards, soft shadow, orange glow)
    25 backgroundsanimated hud_live, 9 dark styles, paper, cream, 10 technology backgrounds (tech:laravel...), 3 ready images; picked per video
    12 card styles + 55 block kindscards, code, terminal, flow, architecture, ER, HTTP, diff, queue, polls, chat window, mini charts, counters
    34 sound effectsassets/sfx, placed automatically at 30% under the voice or pinned by name
    Captions (optional)Arabic in a box, orange highlight on key words and every English word
    Two sizes16:9 and 9:16, both native renders of one scene file
    One commandbuild.py: lint, render, SFX, mux, meta.md, chapters.txt; long videos split into parts
    Safetylint.py (layout, timing, logo, features), check_voice.py (WAV vs scene file), smoke.py (whole-kit test), sfx_preview.py (hear the sounds), all without rendering a video
    Clipsreels.py cuts 30-60 s Reels/Shorts from a long video
    New in v3.0.0scene transitions and camera moves, persistent entities, animated paths (beam), timeline block (tracks), audio events and loudness targets, render cache, storyboard JSON, two-format layout check (RELEASE_NOTES.md)
    Skillsctrlh-video-style (asks before every video) and ctrlh-thumbnail-style

    Quick start for developers (Advanced, optional)

    Only if you want to run the Kit yourself. A regular Claude user does not need this section.

    git clone https://github.com/heshamadel1/ctrl-h.git && cd ctrl-h
    python3 -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt
    cd examples && FMT=wide python3 ../engine/render.py template_howto_scenes frames /tmp/frames 30
    python3 ../engine/build.py my_scenes --vo voice.wav --out out      # a full video
    

    Needs Python 3.13, FFmpeg, Pillow with RAQM (installed for you in Claude's workspace; only developers install them locally). Details: Getting started.

    Documentation

    FileRead it for
    GETTING_STARTED.mdinstall, first frame, first video, Claude skills
    PIPELINE.mdthe one workflow (automated and manual), CLI flags, SFX, captions, formats
    CONFIGURATION.mdevery config key / env variable, priority, themes, backgrounds
    CATALOG.mdevery card, block, theme, background and sound by name and code
    FEATURES.mdevery feature with a picture and its fields
    SOUNDS.mdthe 34 sound effects
    CUSTOMIZE.mdwhat to edit for what; using it for your own channel
    ARCHITECTURE.mdmodules and data flow
    TROUBLESHOOTING.mdcommon errors and fixes
    CONTRIBUTING.mdhow to contribute
    WORKFLOWS.mdStandard Claude vs Advanced agent workflow, optional tools
    STORYBOARD.md, LAYOUT.mdone JSON document for a whole video; Wide/Reel layout check
    CHANGELOG.mdversions and publishing

    With Claude

    Install skills/ctrlh-video-style (and skills/ctrlh-thumbnail-style for thumbnails), then send a script and a WAV. Claude asks, every video: theme, size (16:9 / 9:16), captions, cards, background, sound effects (specific names or automatic), thumbnail.

    License

    Code: MIT (LICENSE). Fonts: SIL Open Font License. The Ctrl + H name and logo are the channel owner's brand assets; do not reuse them without permission.

    • Requires Python Environment: The animation engine requires a compatible Python environment and its dependencies to be installed and configured before use.

    • External Dependencies: Video rendering and audio processing may require additional tools such as FFmpeg and other system-level dependencies.

    • AI Agent Compatibility: The skill follows the open SKILL.md standard, but execution capabilities may vary across Claude Code, Codex CLI, Cursor, Gemini CLI, and other compatible agents.

    • Voiceover Required: Voiceover synchronization workflows require a supplied audio file. Voice generation is not included as a built-in feature and may require an external service.

    • Rendering Performance: Video rendering time depends on video complexity, resolution, effects, and available CPU and memory resources.

    • External Services: Features that rely on external APIs or AI services may require separate accounts, API keys, or usage fees.

    • Output Verification: Generated scripts and assets are not automatically guaranteed to be error-free. Successful rendering and final video quality must be verified by executing and inspecting the output.

    • Customization Required: Custom branding, advanced animation requirements, and specialized visual effects may require editing scene configurations or Python source code.

    • No Automatic Publishing: The kit focuses on video production and rendering. Uploading or publishing videos to YouTube or other platforms is not included unless separately configured.

    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

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    • Free with an Agensi account

    Requires

    GitHub
    Python 3
    pip
    FFmpeg
    Git
    Python

    Permissions

    Environment Variables
    Read Files
    Terminal / Shell
    Browser
    Write Files

    Allowed Hosts

    example.com
    github.com
    api.github.com

    File Scopes

    engine/**
    skills/**
    assets/**
    config/**
    audio/**
    output/**
    *.wav
    *.mp3
    *.mp4

    This skill requires local project file access and terminal execution to run the Python animation engine, process voiceover audio, generate visual assets, and render videos. Python dependencies and FFmpeg must be installed separately where required. Network access and external services are optional and only needed for workflows that use them. The skill should only read and modify files within the designated project workspace and should never expose API keys, credentials, or sensitive environment variables.

    Listed10 days ago
    Updated2 days ago

    What's inside

    Frequently Asked Questions

    Ctrl + H Video Kit is a customizable explainer video production system that combines Claude Skills with a Python animation engine to create professional, voiceover-synchronized animated videos.

    2. Do I need programming experience to use it? Basic familiarity with Python, terminal commands, and project configuration is helpful. The AI skills guide you through the production workflow, but advanced customization may require editing Python code.

    3. Which AI agents are supported? The skills follow the open SKILL.md standard and are designed for compatible agents such as Claude Code, OpenAI Codex CLI, Cursor, VS Code Copilot, and Gemini CLI. Specific capabilities depend on each agent's tool access and execution environment.

    4. Do I need an existing voiceover? For voiceover-synchronized production, yes. You can provide a WAV audio file as the narration source. Voice generation can be handled separately through an external service such as Google AI Studio.

    Yes. The kit supports landscape (16:9) and portrait (9:16) production workflows, allowing you to create content for YouTube and vertical video platforms.

    You need Python 3, pip, Git, and a compatible AI agent. FFmpeg and additional dependencies may also be required depending on the rendering workflow. Check the installation documentation for the complete requirements.

    Yes. You can customize themes, colors, typography, backgrounds, cards, animations, and other visual elements to match your branding and content requirements.

    New in Image & Video