# OpenMotion — All-in-One Studio, AI editing & media fact sheet

[English README](../README.md) · [中文介绍](README.zh-CN.md) · [Asset sources](readme/ASSETS.md)

This page is a source brief for creators, writers and their agents. Use it with the linked implementation documents and actual demo results. It describes the OpenMotion source tree; it is not a record of a public package release.

## Reusable introduction

> OpenMotion turns your local agent into a full-stack video creator. Bring a reference, your brand and your own supported model accounts: study the video, rebuild it in HTML/CSS/JavaScript, generate missing images, clips, music and narration, then edit and render in one Studio. Keep the MP4, editable source and creative decisions for the next cut. BYOA — Bring Your Own Agent. BYOK — Bring Your Own Keys. The local creation software is open source and free to use under Apache-2.0; provider usage has its own costs. Brandkits, Motion Design MD, templates and Skills support explicit reuse today. Connected publishing, collaboration, persistent orchestration, analytics feedback and RAG remain planned.

Editorial framing: lead with **“Reverse engineering for any video.”** Describe the workflow as turning a reference and brand brief into an editable HTML/CSS/JavaScript project, then generating missing media, editing and rendering. **“Your local agent, now a full-stack video creator”** expresses the BYOA workflow; BYOK uses supported accounts and API keys. OpenMotion remains an **All-in-One AI-native Studio for AI editing & Motion**. Local creation software is open source and free to use; agent subscriptions and external model usage retain their own costs.

“Any video” is a creative-workflow ambition, not a guarantee of input retrieval, reconstruction fidelity or audience reach. Model generation depends on configured access, and the public phase-one MCP contract includes eleven tools: nine project tools and two template catalog tools. **Publish to every platform** is planned; MP4 export and a manual publishing handoff are available today. Put product comparisons and upstream acknowledgements in the README's final Acknowledgements section.

## Facts you can cite

| Topic | Current fact | Evidence |
| --- | --- | --- |
| All-in-One Studio | Product scope spanning discovery, assets, storytelling, editing, collaboration, delivery and learning; implementation and roadmap are identified separately. | [Product blueprint](product/studio-vision.md). |
| Super Motion Video Agent | OpenMotion’s proposed creative-agent concept, not a shipped persistent runtime or a substantiated industry-first claim. | [Concept and status](product/studio-vision.md#super-motion-video-agent). |
| Mission | Help everyone become an AI video editor and creator, express their own ideas and build a publishing practice. | [README: mission](../README.md#everyone-can-be-a-creator); product direction, not a reach guarantee. |
| AI editing | The external agent applies editing requests to the project; the user previews and refines clips, timing, text and media in Studio. | [AI editing workflow](../README.md#ai-video-editing), [editing contract](../skills/motion-production/references/edit.md). |
| App | A standalone local application with a Studio editor, timeline and editable projects. | [Product architecture](product/architecture.md), [workspace](project-workspace.md). |
| Agent entry directions | Connect your agent from OpenMotion, or use OpenMotion from your agent. Both work on the same local project; the agent executes in its external client. | [App-first](../README.md#connect-your-agent), [agent-first](../README.md#use-from-your-agent), [adapters](agent-adapters.md). |
| Agent access | Five current Skills; eleven public stdio MCP tools; the pinned native CLI for shell-capable hosts. | [Skills](skills.md), [MCP contract](mcp-phase1-tools.md), [native authoring](../skills/motion-production/references/native.md). |
| Clients | Codex, Claude Code and DeepSeek Harness adapters; WorkBuddy configuration with desktop validation pending; manual local stdio configuration. | [Adapters](agent-adapters.md). |
| Resources | 894 catalog entries: 3 bundled editable templates and 891 external references / registry items. | [Catalog](template-catalog.md), [source snapshot](../web/catalog/hyperframes.json). This is not a count of verified templates. |
| Creative convention | Motion MD v0.1; `MOTION_DESIGN.md` is the filename convention, with the file supplied explicitly to an agent. | [Motion MD](motion-md.md); no automatic importer, parser or validator. |
| Native source | HTML, CSS, JavaScript and media; Hyperframes 0.9.7, GSAP, a React/TypeScript Studio and a Node.js local service. | [Hyperframes integration](hyperframes.md), [package manifest](../package.json). |
| License | Apache License 2.0 for this repository; upstream and font notices retained. | [LICENSE](../LICENSE), [editor](../studio/LICENSE), [templates](../template-library/NOTICE.md). |
| Release | Source checkout and local installation/build paths; npm package publication remains disabled. | [Installation](npm-installation.md), [migration](repository-migration.md). |

## Complete product narrative

Present the Studio as one creative home: **discover → collect → generate → tell the story and cut → review → deliver → learn → create again**. The all-in-one story connects the existing local production workflow with the planned operational and knowledge layers.

- **AI-native architecture:** Super Motion Video Agent is the proposed coordinator for briefs, tasks, tools, models and review. The actual reasoning currently runs in connected external clients.
- **One-stop production and operations:** existing media management, timeline editing and export form the base; shared libraries, creative-task management, scheduling and publishing are coming.
- **Agent assistance and multimodal models:** Skills guide the method, project tools apply changes, and configured models create candidate media for adoption into the timeline.
- **Agent-ready template community:** editable source, changeable parameters, Motion MD, previews and provenance support recreation and personalization. One-click gallery handoff prepares instructions; the connected agent performs the work.
- **Knowledge and feedback:** planned analytics connects outcomes to published versions; personal/team knowledge and RAG provide relevant context; evaluated improvements can become Skills or templates. Current reuse is through explicitly supplied files, not automatic learning.

The [DeepSeek Super-agent research prototype](https://github.com/deepseek-ai/dsh-superagent-prototype) is a design reference for continuing task-oriented work. No runtime is imported by this documentation change. Use “we introduce our Super Motion Video Agent concept”; there is no comparative evidence supporting “the world's first.”

## The timeline and three product pillars

Within the All-in-One Studio, the core experience is an AI-native CapCut: the timeline is where material, story, generated assets, revisions and final output come together. Explain these three pillars with their current status:

1. **Intelligent cuts and story organization.** Agents can apply the user's tone, script and confirmed timecodes to edits; users can arrange and refine material in Studio. Automatic content understanding, irrelevant-passage detection and destination-aware reframing are product directions.
2. **Media generation and search.** Configured Studio models generate images, video, music and voice. A reference-capable image model can produce replacement candidates for a selected image; adopting a replacement preserves its timing and animation. Integrated footage, image, voice, music and Avatar service search is a product direction.
3. **Multiplatform input and output.** Supported local media, text and references retrieved by the host can inform the edit. Current output is MP4 plus editable source, with compositions authored for each destination. Broader platform-link parsing, automatic adaptation and platform presets are product directions.

[OpenChatCut](https://github.com/0xsline/OpenChatCut) informs the explanation of a continuously editable project and an iterative editing lifecycle. It is a conceptual reference; this README revision imports no code or capability from that project.

## Collaboration roadmap — is coming

The planned AI-native video creation and editing space brings people and agents into the same project for editing and motion creation, timeline comments tied to moments or ranges, and collaborative review and revision. The [concept artwork](readme/collaboration-coming.png) is labeled **is coming**; it is generated roadmap artwork, not a screenshot of a shipped interface. Current local project handoffs and revision protection do not constitute shipped timeline comments, collaborative review or simultaneous multiuser editing. No release date is announced.

## Input → Process → Output: reporting boundaries

- **Input:** local media is imported into the project. YouTube / X links are reference inputs for an agent with appropriate retrieval and media tools; the app does not include a universal downloader. Scripts, Markdown, SRT and VTT can be read by the agent; text files are supported by the project editing contract.
- **Process / AI editing:** the agent turns editing intent into cuts, sequencing, text, motion and timing changes in a shared project. Tightening speech requires usable transcripts/timecodes or an agent that can analyze the media. There is no built-in speech-transcription or filler-detection service. Studio's configured Model Gateway can generate image/video/audio assets; generation is not one of the eleven public phase-one MCP tools.
- **Output:** the current export service produces MP4 from the project's dimensions and frame rate. An agent can author portrait, landscape or square versions for different destinations. Dedicated platform export presets, direct publishing and additional video export containers are not implemented. Editable project source remains available for handoff.
- **Iteration:** gather material → establish the story → arrange a rough cut → refine media, motion and sound → preview and revise the same timeline → export a fixed version. Keep the source for the next recut; project revisions and the external agent conversation are distinct.
- **Acceptance:** successful installation, MCP discovery, file rendering, technical verification, normal-speed viewing and user acceptance are separate results. Describe only what your demo actually checked.

## Claims awaiting evidence

“99% recreation” has no defined comparison object, benchmark set or scoring method in this repository. Do not report it as measured code similarity, visual fidelity or template coverage. “Nearly all core Claude Motion templates” likewise has no verified inventory or coverage comparison; the documented catalog provenance is Hyperframes and the individual attributed sources. Broad exposure or viral reach is a creator goal, not a promised outcome.

A future comparison should publish the reference set and versions, permissions, method, outputs, editable projects and failure cases. Planned themed collections and Motion Benchmark / Arena remain proposals.

## Reusable creative resources

The [OpenMotion Library](../library/README.md) contains five collections: optional creative Skills, template starting points, motion component recipes, `MOTION_DESIGN.md` design systems and Brandkits. Entries include a six-profile [Motion Design Remix Skill](../library/skills/motion-design-remix/SKILL.md) with [actual paired previews](motion-design-examples.md), a three-recipe [Presenter Story Skill](../library/skills/presenter-story/SKILL.md), a Brandkit preparation Skill, the three existing template links, a title-reveal recipe, Ink & Water direction and the OpenMotion Brandkit. The presenter section uses three attributed upstream reference/remix pairs, nine catalog clips and the [Presenter Tech](../library/templates/presenter-tech/README.md) demo and locally packaged template. They illustrate real presenter-led editing; the new presenter recipes have not yet produced validated OpenMotion outputs. The six source comparisons are source-backed content adaptations, not an arbitrary-video reconstruction benchmark. These are Library resources, not an automatic marketplace or additional Studio-native template inventory. Presenter Tech has a separate included runner; source, licensing and verification are linked from its local entry. The five core production Skills and eleven MCP tools are unchanged by these Library resources.

## Build a real demo

1. Follow the [source quick start](../README.md#quick-start), or the [host installer](npm-installation.md), in an isolated workspace.
2. Connect the agent and choose the bundled Product Promo example. Record the source version and environment.
3. Ask the agent to create an independent project, replace a headline, adjust its hold and open the same project in Studio.
4. Request an MP4; wait for the job and technical verification, then watch and listen to the result. Keep the editable source.
5. Show the real interaction. Label supplied screenshots as snapshots and the generated cover as brand artwork.

### Prompt: a viral-reference recreation

```text
Use motion-recreate to analyze this accessible reference and rebuild its
hook, pacing and visual structure with my own material. Keep source credits
and asset permissions. Deliver an editable project, an MP4 and a comparison
that identifies both similarities and remaining differences.
```

### Prompt: a talking-head edit

```text
Use this local recording and timed transcript. Propose a concise cut around
one main point, remove the specified irrelevant passages, add readable
titles and supporting visuals, and use my supplied music. Keep the meaning
and the speaker's intent. Preview the edit; deliver the project and MP4.
```

### Prompt: an article or introduction video

```text
Read the README, this fact sheet and the linked product documentation.
Lead with “The All-in-One AI-native Studio for AI editing & Motion.” Present
AI-native CapCut as the core timeline experience, with Claude Motion informing
the motion-creation direction. Introduce our Super Motion Video Agent concept
and the planned operations and knowledge loop: publish, collect outcomes,
retrieve personal/team context, evaluate improvements and reuse Skills. Explain intelligent cuts and story organization,
media generation and search, and multiplatform input/output with their status.
Show both entry directions: connect an agent from OpenMotion, or use OpenMotion
inside an agent through Skills / MCP / CLI.
Use Input → Process → Output: show a real editing request, the agent’s project
changes, Studio preview and a revision. Use the supplied artwork and actual UI
screenshots, with attribution.
Separate implementation, product direction and your own assessment.
Include a source-based getting-started path and the Apache 2.0 license.
Do not invent performance, coverage, adoption or audience-growth figures.
```

## Visual assets

| Asset | Intended use |
| --- | --- |
| [Cover PNG](readme/cover.png) | Full-resolution generated brand artwork. |
| [Animated cover GIF](readme/cover-motion.gif) | Short looping brand motion for a README; not a recording of product behavior. |
| [1280 × 640 social preview](readme/social-preview.jpg) | Static GitHub/social sharing cover. |
| [Three-mode product tour](product-tour.md) | Six captures: Codex's embedded Viewer, standalone Studio, Claude Code Skill discovery, native CLI checks, Connect Agent and a single-template handoff. Codex host chrome and model-completed adaptation are not captured. Terminal images replay actual recorded output. |
| [Workspace](readme/workspace.jpg) / [editor](readme/editor.jpg) | Historical Studio snapshots at commit `7aee372`; retained for provenance. The README now uses the three-mode tour. |
| [Seven creative examples](examples.md) / [中文简报](examples.zh-CN.md) | Seven curated external reference excerpts, each with an animated and static preview, attribution and an actionable creative brief. These are creative inputs, not OpenMotion-generated demos or new bundled templates. |
| [18 animated catalog picks](../README.md#motion-gallery) | Six groups of actual reference/registry video excerpts: product, launch, type, light/material, spatial and generative motion. Default previews are GIFs with source credit; these do not represent additional bundled templates. [Manifest](readme/gallery/manifest.json). |
| [Library banner](readme/library.png) / [community banner](readme/community.png) | Generated artwork in the same painterly family as the cover. |
| [Collaboration concept](readme/collaboration-coming.png) | Planned human–agent editing and timeline review, labeled **is coming**; use only as a concept preview. |
| [Stay tuned GIF](readme/stay-tuned.gif) | Generated Star invitation, linked to the GitHub repository; release updates require Watch configuration. |
| [Brandkit](../library/brandkits/openmotion/BRANDKIT.md) / [motion design system](../library/design-systems/ink-water/MOTION_DESIGN.md) | Reusable identity and creative direction; supply explicitly to your agent. |

See [ASSETS.md](readme/ASSETS.md) for the image prompt, animation recipe, catalog-image provenance and third-party notices. Referenced films and example brands belong to their respective creators; the application license does not relicense them.