This Just Killed AI Video Editing With GPT 6 Astra
Key Takeaway
By routing GPT-6 Astra through Claude Code via CLI Proxy API and using Hyperframes skills, developers can build fully automated, interactive 30-second product demo videos without paying per-token API charges.
Highlights
Running GPT-6 Astra inside Claude Code using CLI Proxy API avoids direct OpenAI API token costs by leveraging an existing Codex subscription.
A custom script or proxy server connects Claude Code to third-party models by overriding the local base URL setting.
Drafting an interactive video plan on a local web server with comment capability prevents wasted render time and unnecessary API token usage.
Installing Hyperframes skills allows AI coding agents to control browser camera movements, zooms, and screen captures frame-by-frame.
Automating background music selection involves searching Mixkit for free audio categories aligned with product visual transition timing.
Timeline
Configuring GPT-6 Astra inside Claude Code using CLI Proxy API
Creating video content requires substantial token consumption. Utilizing CLI Proxy API allows developers to bypass standard API token charges by leveraging existing subscription tiers. Setting the base URL in Claude Code points internal requests to the local proxy server, making GPT-6 Astra available directly inside the editor environment.
- Direct OpenAI API usage for AI video generation incurs high token costs.
- CLI Proxy API creates a local server on the host machine to route Claude Code requests through an existing Codex plan.
- Overriding the base URL in Claude Code directs traffic away from Anthropic servers to local models.
Interactive shot planning and web-based feedback
Before generating raw video frames, the agent maps out shot timing, aspect ratios, and camera focal points based on direct product URLs and reference videos like Linear or Anthropic launches. Displaying this plan on a local web page allows creators to leave exact feedback and request structural updates before locking the storyboard.
- Unplanned video generation leads to hours of wasted rendering time.
- Hyperframes skills enable AI agents to plan camera paths and display interactive blueprints directly in the browser.
- Locking the final storyboard prevents unwanted modifications during subsequent shot production.
Luma AI Ray 3.2 overview
Generative video production relies heavily on rendering efficiency and consistency. Ray 3.2 improves prompt accuracy while maintaining frame-to-frame stability across cinematic and styled renders. Built-in editing tools let users modify post-render frames directly.
- Luma AI Ray 3.2 renders native 1080p HDR video at four times the speed and one-third the cost of Ray 3.0.
- Keyframe controls and modify tools allow direct fixes to rendered clips without starting over.
Generating and refining individual video shots
Generating each scene as an independent clip prevents cascading rendering errors across the sequence. Reviewing early shots identifies camera shake, incorrect zoom levels, or layout bugs so adjustments can be made before full composition.
- Rendering shots separately isolates errors and speeds up revisions.
- Reviewing the initial shot catches movement and zoom issues early in the pipeline.
Combining scenes and adding synchronized audio
After scene capture, Claude Code stitches the individual video segments into a single cohesive video file. A dedicated audio selection skill evaluates transition pace and product style, automatically querying royalty-free music libraries like Mixkit for fitting backing tracks.
- Custom skills analyze visual transition timing to select matching audio from Mixkit.
- Automated video checks eliminate small visual artifacts before final export.
Description
AI video editing tools inside Claude Code: we ran GPT-6 Astra on a real product demo. Visit https://lumalabs.ai/ailabs30 to try Luma today No hunting for the best ai video editor, no stacking random ai tools. Astra plans and films the shots, Claude Code runs every tool. Community with All Resources: https://ailabspro.io/?v=p565AqELA8M The Roundup, our daily newsletter covering the AI stories that matter. Join now: https://www.theroundup.so/ What we cover in this video - The full ai video editing workflow we follow, from planning the shots to filming them to adding the music - Why we run GPT-6 Astra inside Claude Code, so Astra gives the creative direction and Claude Code runs the tools - Why Astra over Claude's own models: it has the better design direction, and it came out ahead on design in our Fable 5.1 comparison video - How to do the same thing with Codex or any other coding agent if you don't want Claude Code Setting up Astra in Claude Code - Why we don't recommend paying for OpenAI's API for this, since video editing by ai burns a lot of tokens and gets expensive fast - Using CLI Proxy API to run Astra through the Codex subscription you already pay for - Installing it with one brew command, then letting your agent handle the settings and the API key - Changing the base URL so Claude Code talks to the local server instead of Anthropic - Installing the video tools and skills the agent needs before any video editing with ai can start Planning the video before you film anything - The product we filmed: an artistic clothing store built from the app template in our community - Giving the agent the address of the running app so it can open it and plan around what's actually there - Choosing the length and the layout, 30 seconds in 16:9 for product demos, or 9:16 for short-form - Asking for music with no voiceover as part of the plan - Getting the plan rendered in your browser with a comment box, so you can see the shots instead of reading them - Using Anthropic's and Linear's launch videos as references, which is what tells the agent how the camera should move and where it should zoom - Locking the plan once you're happy, so the direction is final Framing and filming each shot - Why we record every shot separately, so a bad shot only costs you that shot - Recording at full-screen resolution and turning each one into a clip you can actually watch - Reviewing the first shot before the agent generates the rest, so one mistake doesn't repeat through the whole video - Fixing shaky clips and zooms that drifted from the plan - This is the stage most ai for video editing attempts skip, and it's why they end up redoing everything Polishing and combining the clips - Telling the agent how the camera should move, where it should zoom and when the typing should happen - Combining the separate clips into one video that flows from shot to shot - Finding the output in the working folder and telling the agent exactly where a problem happens - Running the built-in checks that catch the small issues before you call it done - No ai video editing app involved, the agent drives the tools itself Adding the music - Why music is the part that keeps a viewer hooked through every shot - What happens when you just ask for music with no guidance, and why the track usually doesn't fit - The skill we built that reads the product's style from the plan and counts how often the video cuts - Matching the beat to the cut and pulling free tracks from Mixkit in categories like calm and piano - Running it with a slash command so it finds the track and renders a new version with the music in Who this is for - Anyone building a product who needs a demo video that gets attention without making people install anything - Non-technical builders who want ai editing video work done for them, without learning an editor - People who have tried chatgpt video editing in a chat window and want the same model driving real tools instead - Anyone comparing claude ai video editing setups and trying to work out which model should lead - Anyone doing video editing using ai who keeps getting camera moves and ai video animation that don't match the plan 00:00 Intro 01:32 Setting up Astra 04:14 Planning the video 06:24 Sponsor 07:21 Creating the shots 08:41 Combining shots 09:39 Adding music #aiVideoEditing #aiForVideoEditing #chatgptVideoEditing #aiVideoAnimation #gpt6Astra #ClaudeCode #Fable51