# Vibe Vision Video

> Open-source Remotion video templates and an Agent Skill that let coding agents render, check and deliver reusable films from your photos and copy.

Vibe Vision Video (vvv) is an open-source set of editable Remotion video templates and an Agent Skill for coding agents such as Claude Code and Codex. A user describes a film in one sentence, and the agent picks a template, places photos, writes copy, renders a sample, checks it and delivers the film with its editable project. The current template library focuses on wedding films. The latest release listed is v0.1.1, and the README describes the project as an early 0.1 release.

## What It Is

vvv packages video templates so an agent does not have to redesign and debug the same video code for every film. Scene structure, camera moves, 16:9 and 9:16 layouts, beat-planned timing and render checks are built once. Each new film changes only photos, names and dates, copy lines, music and format. Templates include Album, Comedy, Doodle, DV Story and Starter.

## How the workflow runs

The `vvv.py describe` command tells the agent a template's photo slots, copy fields and settings, and `validate` catches bad inputs before rendering. The agent changes template inputs first and edits React/TypeScript source only where the brief needs it. Rendering runs locally and requires Python 3.9+, Node.js 26 and npm. `vvv check` measures format, audio, still stretches and beat alignment, then lists what only a viewer can judge.

## Recreate a reference film

Given a local reference video plus photos, a logo, copy and music, the agent measures sections in beats, shot lengths, still frames and beat alignment. It then rebuilds the film on the closest template or a new composition, generates missing artwork and checks the result against the reference. The reference is studied, not shipped. Video and voice are not generated. Image generation works best with an image-capable agent; Claude Code can use Codex's image generation through a bundled bridge when both are installed.

## Agents, MCP and community

Claude Code installs it as a plugin, while Codex, Cursor, Gemini CLI, Copilot and other agents read it from the shared skills folder. A hosted MCP server lets clients browse templates, read photo slots and copy fields, and fetch template packages and guides. Templates and finished projects can be packed as a single .vvv file, and community templates can be submitted for owner review.

## Current Status

The README says it is verified end to end on macOS ARM64 with Claude Code and Codex, while WorkBuddy and other platforms are less tested. Code, docs and templates are MIT licensed, while fonts and music keep their own terms.

## Features
- Reusable Remotion video templates with documented input contracts
- Agent Skill for Claude Code, Codex and 25+ agents
- Recreate a reference video as an editable project
- Render checks for format, audio, still frames and beat sync
- 16:9 and 9:16 layouts
- Hosted MCP server for browsing templates and guides
- Portable .vvv packaging for handoff
- Local FFmpeg footage helper
- Community template submission

## Integrations
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, WorkBuddy, OpenCode, Windsurf, Cline, Roo Code, Remotion, FFmpeg, MCP

## Platforms
MACOS, LINUX, WEB, API, VSC_EXTENSION, DEVELOPER_SDK, CLI

## Pricing
Open Source

## Version
v0.1.1

## Links
- Website: https://vibevision.video
- Documentation: https://github.com/actionow-ai/vvv/tree/main/docs
- Repository: https://github.com/actionow-ai/vvv
- EveryDev.ai: https://www.everydev.ai/tools/vibe-vision-video