Kling AI Prompt Generator - Claude Skill

A Claude skill for generating professional Kling AI video prompts. Works with Claude.ai, Claude Desktop, and Claude Code. Supports Kling 3.0, Avatar 2.0 (talking avatars), Multi-Shot Storyboards, Character LoRA, Draft Mode, and Motion Control workflows.

What is this?

A skill for Claude that helps you write professional Kling AI video prompts - structured, model-aware, and ready to paste. Instead of guessing what works, Claude asks clarifying questions and builds the prompt following Kling's cinematic conventions.

Works with Claude.ai (web), Claude Desktop, and Claude Code - all three support the same skill format via the Customize > Skills panel.

Developed and used daily in production by Maciej Dzierżek - consultant, trainer, and creator from Poland who works with Kling AI professionally.

Try the interactive web version (no installation needed): maciejdzierzek.com/narzedzia/generator-kling-ai

Demo - one-shot prompt from this skill

The video below was generated in Kling AI VIDEO 3.0 from a single prompt built by this skill - no retries, no edits, no post-production. It demonstrates four features the skill actively pushes you toward: 4-shot storyboard, character consistency across shots, explicit camera direction, and native audio with speaker attribution.

https://github.com/user-attachments/assets/87a57287-5d9b-4c43-b430-46860222951d

If your client doesn't render the player above, watch the demo here.

A creator types a prompt → match cut into the world the prompt describes (a samurai in neon Tokyo rain) → close-up of the same samurai → pull back reveal it was all playing on the creator's monitor. 12 seconds, 720p, one generation.

Features

• 6 generation modes: Image-to-Video, Text-to-Video, Multi-Shot Storyboard, Avatar 2.0 (talking avatars), Motion Control, Canvas Agent (one-click storyboard)
• Model-aware: Covers the current Kling lineup - VIDEO 3.0, VIDEO 3.0 Turbo, VIDEO 3.0 Omni, VIDEO 3.0 Motion Control, Avatar 2.0, 2.6, 2.5 Turbo - with model-specific prompt length and element limits
• Avatar 2.0 workflow: Generate a talking avatar from one photo + audio - up to 5 minutes of continuous output
• Element Reference: Kling's official mechanism for cross-shot character/scene consistency. Multi-image references on VIDEO 3.0, plus video references on Omni
• Voice Control: Custom voice creation with [Character] @VoiceName syntax (VIDEO 2.6 / 3.0, EN+ZH bidirectional)
• Canvas Agent guidance: One-click storyboard, multi-angle expansion, reverse prompt suggestions, batch generation
• Camera movement vocabulary: 10 named movements (static, dolly, orbit, tracking, crane, POV...) with speed modifiers
• Motion endpoint pattern: Adds "returns to starting position" / "then settles" to prevent the stuck-at-99% generation hang
• Audio prompts: Speaker attribution syntax for lip-synced dialogue (5 languages on VIDEO 3.0: EN, ZH, JA, KO, ES)
• Sequential action support: "First / then / finally" action structure for multi-step scenes
• Native 4K guidance: 30 credits/sec in-app on VIDEO 3.0 and Omni (since 2026-04-23); not available on Turbo
• Video editing: prompt patterns for editing an existing clip on VIDEO 3.0 Omni - object swap, background/subject replacement, targeted attribute change, style transfer with trajectory lock (3-15s input, 4K in and out)
• API rate card in Units: per-second cost by model x resolution x native-audio x video-input, so you can price a deliverable before generating it
• 10+ ready-to-use templates: Product, portrait, atmospheric loop, commercial, talking head, action scene, Avatar presenter/explainer/singer
• Always outputs English prompts: Regardless of the language you describe your scene in
• Sourced from official Kling release notes: every major fact in the skill is verifiable against kling.ai/release-note - URLs cited inline

Installation

Claude.ai and Claude Desktop (all plans, including free)

1. Download kling-ai-prompt-generator-skill.zip from this repository (click → Download raw file)
2. In Claude.ai or Claude Desktop: go to Customize > Skills → click "+" → "Upload a skill"
3. Upload the ZIP file and toggle the skill on

Claude will automatically use the skill when you describe a Kling AI video task. Available on Free, Pro, Max, Team, and Enterprise plans.

Note: Skills require code execution to be enabled. If you don't see the Skills section, go to Settings > Capabilities and enable "Code execution and file creation" first.

Claude Code (CLI) - via plugin marketplace

/plugin marketplace add maciejdzierzek/claude-plugins
/plugin install kling-ai-prompt-generator@maciejdzierzek

One marketplace, all my skills - see maciejdzierzek/claude-plugins. No download needed.

If you added this repository as a marketplace before August 2026, switch to the command above. Claude Code registers one marketplace per name, so the old per-repo catalogues (which all used the same name) replaced each other when you installed more than one skill. This repo's catalogue still resolves, under the name maciejdzierzek-kling, but it is no longer the recommended path.

Claude Code (CLI) - manual install

mkdir -p ~/.claude/skills/kling-ai
cp skills/kling-ai/SKILL.md ~/.claude/skills/kling-ai/SKILL.md

For project-specific use:

mkdir -p /your-project/.claude/skills/kling-ai
cp skills/kling-ai/SKILL.md /your-project/.claude/skills/kling-ai/SKILL.md

Reload Claude Code. The skill activates automatically.

Interactive web generator (no installation)

Use directly at: maciejdzierzek.com/narzedzia/generator-kling-ai

No API key, no registration, no data sent to servers. Works entirely in your browser.

Usage

Describe what you want to create - Claude will identify the right mode and build the prompt:

Create a Kling AI prompt for an image-to-video animation of my product photo.
The product is a perfume bottle and I want subtle light reflections moving.

I need a text-to-video prompt for a cinematic portrait of a woman in a cafe.
She should slowly turn toward the camera. Using Kling 3.0.

For Avatar 2.0 (talking avatars):

I have a portrait photo and a 45-second narration audio file. Build me an Avatar 2.0
prompt for a corporate explainer video with a calm, professional tone.

Use a template as starting point:

Load the "Product Commercial" Kling AI template and adapt it for a matte black watch.

For multi-shot sequences:

I want to create a 4-shot storyboard in Kling 3.0 for a product launch video.
The product is headphones. Approximate total duration 12 seconds.

Prompt Structure

Image-to-Video

[MOTION DESCRIPTION - what moves, not what exists].
[STATIC ELEMENTS: what must remain fixed].
Camera: [MOVEMENT]. Duration: [5/10/15]s. Aspect ratio: [16:9/9:16/1:1].

Text-to-Video

Setting: [ENVIRONMENT + TIME OF DAY + ATMOSPHERE].
Subject: [CHARACTER/OBJECT DESCRIPTION].
Action: First [INITIAL STATE], then [MAIN ACTION], finally [RESOLUTION].
Camera: [MOVEMENT + LENS]. Style: [VISUAL KEYWORDS].
[AUDIO if using V3/O3/2.6]. Duration: [5-15]s.

Multi-Shot Storyboard (V3/O3 only)

Shot 1 ([Xs]): [Wide establishing shot]. Camera: static.
Shot 2 ([Xs]): [Medium shot, action begins]. Camera: slow push in.
Shot 3 ([Xs]): [Close-up, reaction or detail]. Camera: static.
Shot 4 ([Xs]): [Resolution]. Camera: tracking / pull back.

Audio Speaker Attribution (V3, O3, 2.6)

[Speaker: Character Name] "[dialogue]" in a [warm/confident/excited] [male/female] voice with [accent].
Add [sound: footsteps / rain / door closing] when [action].
Background ambient: [environment description].

Examples

See examples/ for ready-to-use prompts with notes on expected results:

• examples/image-to-video.md - Product animation, portrait animation, nature loop, logo animation
• examples/text-to-video.md - Cinematic portrait, product commercial, atmospheric loop, talking head with audio
• examples/motion-control.md - Orbit, push in, Ken Burns pan - with camera path setup notes

Models

"Kling 3.0" is a series, not one model. The official 2026-01-31 release note distinguishes VIDEO 3.0 (upgrade from 2.6) and VIDEO 3.0 Omni (upgrade from O1); the 2026-06-17 release note added VIDEO 3.0 Turbo. All share the unified multimodal architecture; Omni adds video element reference, element voice control and video editing; Turbo trades the 4K tier for lower per-second cost and the strongest lip sync in the family. For long-form talking-head content (>15s), use Avatar 2.0 - it's purpose-built and supports up to 5 minutes. Native 4K is a separate 2026-04-23 release available on the VIDEO 3.0 series at 30 credits/sec.

About the Author

Built by Maciej Dzierżek - consultant, trainer, and creator based in Poland. He helps businesses implement AI in their workflows through training, consulting, and hands-on tools.

• Website: maciejdzierzek.com
• All AI tools: maciejdzierzek.com/narzedzia

License

MIT - use freely, attribution appreciated. See LICENSE.