Copy the VIX connector URL
You'll paste it in the next step.
https://mcp.vixsbl.com/mcpGenerate, edit, and compose media straight from your prompts. Nothing to install, no API key to manage.
You'll paste it in the next step.
https://mcp.vixsbl.com/mcpThe button lands you on the add-connector screen. Name it VIX and paste the URL.
Approve VIX on the consent screen. You're set, ask Claude to make something.
Ask for what you want. Your agent picks the operations and hands the file back.
Nothing yet
Some edits take a few steps. Your agent runs them in order and hands back the finished file.
“Transcribe this interview, pull the strongest 30 seconds, and burn captions into the cut.”
“Animate these four product stills into slow push-ins and stitch them into one 20 second video.”
“Cut this product out of the photo, take it up to 4K, and hand me a transparent PNG.”
“Read this script in a warm voice and lay it under my b-roll at the right level.”
“Dub this ad into Spanish and keep the captions in sync.”
“Take the music bed off this clip and give me a clean voice track.”
Add it as a custom connector in Claude or ChatGPT and paste the URL. Nothing to install and no key to manage. Terminal agents like Claude Code and Codex can use the CLI instead.
Claude on web, desktop, and mobile, ChatGPT with Developer mode on, and any MCP-compatible client via the connector URL. Terminal agents like Claude Code, Codex, Hermes, and OpenClaw use the published CLI.
Yes. Every upload and generated output comes back as a reusable handle, so your agent can feed one result into the next step or pick it back up later in the conversation.
Install the VIX skill with npx skills add soberbunnylabs/vix-skills. It teaches your agent when a task calls for VIX and comes with short lessons on chaining operations into a finished result.
No. You sign in with your VIX account through Claude's OAuth flow. The only thing to paste is the URL.
Images, video, audio, and edits, plus multi-step pipelines that combine them into a finished result.
Runs draw from your VIX plan's usage budget, shared across the app, the CLI, and every connected agent. No per-call credits.
Runs are asynchronous. Your agent submits a run and checks back, so nothing hangs while a render completes.
You approve the connection on a consent screen, and the connection only ever sees an operation request and returns a result.
Yes. The CLI is built for terminal and headless agents, and is published as @soberbunnylabs/vix-cli.
One paste and your agent is wired into VIX.
https://mcp.vixsbl.com/mcp