Turn Claude Code Into a Video Production Studio
Production-ready AI skills and workflows that automate cinematic video creation with Google Flow. From script to YouTube in minutes, not days.
Open source · MIT License · 32 skills · 10+ visual styles · Free forever
# Create a cinematic documentary in one command
> /gla:create-project "The Lost City of Petra"
> /gla:gen-refs # generate character references
> /gla:gen-images # create scene images with consistency
> /gla:gen-videos # animate every scene
> /gla:concat # merge into final video
> /gla:youtube-upload # publish to YouTube
Tired of Wrestling With AI Video?
Creating multi-scene AI videos manually is painful. You spend more time on workflow than creativity.
Hours of Manual Prompting
Writing individual prompts for each scene, character, and angle. Copy-pasting between sessions. Losing context mid-project.
Inconsistent Characters
Every generated scene looks different. Characters change appearance between shots. No visual continuity across your video.
Complex Post-Production
Juggling ffmpeg commands, scene transitions, TTS narration, music layers, and YouTube uploads across disconnected tools.
Meanwhile, other creators are shipping cinematic videos with a single command...
FlowKit: Your AI Video Production Team
Everything you need to create professional multi-scene videos directly from Claude Code.
Pre-built workflows for video creation, TTS, music, thumbnails, and YouTube publishing
Full REST API covering projects, scenes, characters, requests, and direct generation
3D Pixar, anime, realistic, stop-motion, Minecraft, oil painting, and more built-in
32 Skills That Do the Heavy Lifting
Each skill is a battle-tested workflow. Type the command in Claude Code and watch it execute.
Core Pipeline
/gla:create-projectInteractive project setup with entities, scenes, and visual style
/gla:gen-refsGenerate reference images for consistent characters and locations
/gla:gen-imagesCreate scene images with character references applied
/gla:gen-videosAnimate every scene image into cinematic video
/gla:concatDownload and merge all scenes into final video output
Advanced Video
/gla:gen-chain-videosSmooth scene transitions with start+end frame chaining
/gla:insert-sceneInsert cutaways, close-ups, and multi-angle shots into chains
/gla:creative-mixAI-suggested cinematic techniques for your story
/gla:review-videoClaude Vision quality review before upscaling
/gla:add-materialApply visual styles: 3D Pixar, anime, realistic, and more
TTS & Music
/gla:gen-narratorGenerate narrator text + TTS for all scenes (600+ languages)
/gla:gen-tts-templateCreate consistent voice templates for narration
/gla:gen-musicGenerate background music via Suno integration
/gla:concat-fit-narratorTrim scenes to fit narrator duration with text overlays
/gla:import-voiceImport existing voice as template for consistency
YouTube Pipeline
/gla:youtube-uploadUpload to YouTube with scheduling and rule validation
/gla:youtube-seoGenerate SEO-optimized titles, descriptions, and tags
/gla:thumbnailGenerate 4 YouTube-optimized thumbnail variants
/gla:brand-logoApply channel logo watermark to video and thumbnails
/gla:camera-guideCinematic camera angles, movements, and lighting reference
Project Management
/gla:pipelineFull pipeline orchestration from start to finish
/gla:statusComplete project dashboard with recommended next action
/gla:monitorReal-time monitoring of running pipeline operations
/gla:researchFact-check and research before scripting for accuracy
/gla:refresh-urlsRefresh expired GCS signed URLs for all scene assets
From Script to YouTube in 6 Commands
Each command handles an entire stage of the video production pipeline. No manual steps.
/gla:create-projectCreate Your Project
Define your story, characters, locations, and scenes. FlowKit sets up the entire project structure with consistent entity references.
Supports dual-orientation (16:9 + 9:16 Shorts) from one project
/gla:gen-refsGenerate References
Create reference images for every character, location, and asset. These ensure visual consistency across all scenes.
Characters maintain their appearance throughout the entire video
/gla:gen-imagesGenerate Scene Images
Each scene image is generated with character references automatically applied. Choose from 10+ visual styles.
3D Pixar, anime, realistic, stop-motion, oil painting, and more
/gla:gen-videosAnimate to Video
Transform scene images into cinematic video clips. Supports scene chaining for smooth transitions between shots.
Optional: add narration, music, and text overlays
/gla:concatMerge & Finalize
Concatenate all scenes into a polished final video with TTS narration, background music, and 4K upscaling.
Auto-trims scenes to match narrator timing
/gla:youtube-uploadPublish to YouTube
Generate SEO metadata, thumbnails with brand watermark, and upload directly to YouTube with scheduling support.
Supports both Shorts (vertical) and long-form (horizontal)
Get Started in 2 Minutes
FlowKit is open source and free forever. Clone, install, and start creating videos.
Clone the repository
$ git clone https://github.com/crisng95/flowkit.git
$ cd flowkit
Install & start the agent
$ pip install -r requirements.txt
$ python agent/main.py
Start creating in Claude Code
# Open Claude Code in your project directory
> /gla:create-project "My First AI Video"
> /gla:pipeline # full automated pipeline
What Creators Are Saying
Real feedback from video creators using FlowKit in production.
“I used to spend 4-5 hours manually prompting each scene. With FlowKit, my entire 12-scene documentary was done in under 30 minutes. The character consistency alone is worth 10x the price.”
Alex T.
Discord
“The scene chaining feature is insane. Smooth transitions between every shot without me touching ffmpeg once. This is how AI video should work.”
Maria S.
“From create-project to YouTube upload in one session. I shipped 3 videos yesterday that would have taken me a week before FlowKit.”
David L.
Discord
“The dual-orientation output is a game-changer. Same project generates both my YouTube long-form and Shorts. No extra work, just works.”
Sarah K.
“Reference images solved my biggest pain point. My characters actually look the same across all scenes now. No more identity drift.”
James R.
Discord
“The TTS narrator feature with 600+ languages opened up markets I couldn't reach before. Now I publish in English, Vietnamese, and Spanish from the same project.”
Nina P.