mcpField
B
BeatDesign

BeatDesign

Local
github.com/BeatAPIMedia & Design274Updated 1w ago

About BeatDesign

BeatDesign is a local-first workspace for AI image and video creation, aimed at creators who want to build and edit media workflows in one project. Its Canvas, browser-based video editor, shared assets, and agent access all work on the same local project.

Key features

  1. 01Infinite Canvas for media workflows
  2. 02Browser-based timeline video editor
  3. 03Shared assets across workspace modes
  4. 04Local preview and MP4 export
  5. 05English, Chinese, and Japanese UI

Use cases

  1. 01Branch and compare image or video generations
  2. 02Arrange clips and audio on a timeline
  3. 03Add image overlays and styled SRT captions
  4. 04Ask an agent to update a project

MCP support

BeatDesign provides local stdio and Streamable HTTP MCP transports. MCP clients can access tools for projects, assets, Canvas, generation, and Editor operations; agent changes appear in the browser workspace.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository
Packages
mcpbbeatdesign-0.2.1.mcpb

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

BeatDesign infinite Canvas and local video Editor

README image

BeatDesign

Your local Higgsfield alternative.
Build image and video workflows on an infinite Canvas, finish them in a browser-native Editor, and let any MCP-capable Agent work in the same Project.

Website · Quick start · MCP · Product status

English · 简体中文 · 日本語

BeatDesign is an independent, open-source, local-first workspace for AI image and video creation. It is built for people who want the connected creative flow of tools such as Higgsfield, with a workspace they can run, inspect, and extend on their own machine.

Latest release — v0.2.5 · October 2026

Available in GitHub Release v0.2.5 and @beatapi/[email protected]. WorkBuddy marketplace review and availability are tracked separately.

For an existing source checkout, stop BeatDesign, run git pull --ff-only, pnpm install --frozen-lockfile, pnpm db:push, then restart pnpm dev:agent (or your usual UI/MCP commands). Keep your local data directory; no reinstall or project reset is needed.

  • Project settings: edit names and descriptions, upload or select a cover, and reset to an automatic image or local video poster. A manual cover stays selected after later imports and generations. Open the settings icon on a project card or workspace header.
  • Consistent generation versions: Canvas previews and Take selection agree on the latest successful output; manual selections and concrete timeline Assets are preserved.
  • Faster project entry: choosing Canvas or Studio creates one project and opens it directly, with retry protection.
  • Reliable audio export: MCP timeline rendering preserves clip gain across sequential shots, with peak protection for overlapping audio.
  • 30 MCP tools: Agents can update project settings through the same revision-checked command path as the UI. Canvas undo/redo can save an intentionally empty canvas; Windows test portability and narrow-screen layout are improved.

Already available: tail-frame continuation, Canvas-to-Editor handoff, image overlays, styled SRT captions, browser-native MP4 export, and MCP timeline rendering. Browser export needs no system FFmpeg; MCP frame extraction needs ffmpeg, and MCP timeline rendering also needs ffprobe.

Change details · Project settings · Current scope and roadmap

Current source (unreleased) adds persistent Canvas groups and arrangement, Asset folders, exact-version Elements and reusable local workflow templates, with 38 MCP tools. These changes are not in the v0.2.5 npm package. Organization guide.

Two creative modes, one Project

Infinite Canvas

Drop in prompts, images, video, and audio. Connect nodes, branch ideas, compare generations, reuse outputs, and keep the full visual workflow inside one project.

Video Editor

Turn Canvas results into a local timeline. Trim, split, move, mix audio, place and transform image overlays, style SRT captions, create alternate AI Takes, preview, and export MP4 in the browser.

Studio provides a focused generation surface. Assets keeps every imported and generated file available across Canvas, Editor, and MCP.

Why BeatDesign

  • Local by default. Projects, media, Canvas state, timelines, and generation history stay in your local workspace.
  • Canvas and Editor stay connected. A generated result becomes a reusable project Asset instead of disappearing inside one tool.
  • Bring your own Agent. Codex, Claude Code, Cursor, and other MCP hosts can inspect and update the same Project you see in the browser.
  • Bring your own generation access. BeatAPI is the built-in provider. Local importing, arranging, editing, previewing, and exporting do not require an API key.
  • Fork-friendly foundations. Provider adapters, project storage, and the command layer are explicit extension points.

From idea to finished video

Prompt or local media
        ↓
Infinite Canvas ─── connect, generate, branch, compare
        ↓
Shared Assets ───── images, video, audio, extracted clips
        ↓
Video Editor ────── trim, arrange, preview, export MP4
        ↑
Any MCP Agent ───── works in the same local Project

Quick start

Requirements include Node.js 22+, pnpm 10+, and a current Chrome browser on macOS or Windows.

pnpm install
pnpm db:push
pnpm dev

Open http://127.0.0.1:3020.

Add your own BeatAPI API key only when you want to generate or analyze media, or use AI redo. Selecting a local file does not upload it to a provider.

Choose how to run BeatDesign
CommandStarts
pnpm devThe visual workspace
pnpm dev:agentThe visual workspace and local HTTP MCP endpoint
pnpm mcpThe stdio MCP server for coding Agents

Use BeatDesign with an Agent

Current source exposes 38 local MCP tools (v0.2.5 packages expose 30) for Skills, Projects, Assets, Canvas, generation, and Editor operations, including built-in Skill discovery and authoritative MP4 timeline rendering. Agent changes use the same project services and become visible in the browser workspace.

After connecting your MCP host, you can ask:

Open my latest BeatDesign project, add these clips to a timeline, place the logo overlay, adjust the subtitles, and leave the Editor open for review.

Platform compatibility

Coding Agent / platformStatusQuick setup
Claude Code✅ SupportedPlugin or MCP config
Codex✅ SupportedPlugin setup
ZCode✅ SupportedMCP config
OpenCode✅ SupportedMCP config
Cursor✅ SupportedMCP config
Windsurf✅ SupportedMCP config
VS Code + GitHub Copilot✅ SupportedMCP config
Cline / Roo Code✅ SupportedMCP config
Qwen Code✅ SupportedMCP config
Gemini CLI✅ SupportedMCP config
Hermes Agent✅ SupportedMCP config
Kiro✅ SupportedMCP config
Trae✅ SupportedMCP config
WorkBuddy✅ SupportedConnector setup
QwenWork✅ SupportedHTTP connector
Doubao Work✅ SupportedHTTP connector

Open the full Agent integration guide →

What works today

  • Image and video generation plus Standard and Deep video analysis.
  • Image, video, audio, generation, timeline, and text nodes on the Canvas.
  • Project Assets shared across Studio, Canvas, Editor, and MCP.
  • Non-destructive timeline editing with image, video, audio, movable image overlays, and styled SRT caption tracks.
  • Local preview and browser-side H.264/AAC MP4 export with WebCodecs and Mediabunny.
  • English, Chinese, and Japanese interfaces.
  • Local stdio and Streamable HTTP MCP transports.

BeatDesign currently focuses on local, short-form creative workflows. Hosted collaboration, multiple named timelines, advanced transitions, speed controls, waveforms, and desktop packaging remain outside the current release.

See the complete shipped and planned scope →

Project guides

License

BeatDesign is licensed under Apache License 2.0. BeatAPI and BeatDesign trademarks, third-party model access, and bundled third-party components remain subject to their respective terms. See third-party notices and trademark policy.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.