原型 · 预览版:收录官方注册中心里的全部服务器,每 30 分钟同步一次。评价和账户功能将在后续阶段推出。
PopNx MCP为使用 AI 的人准备的 MCP 服务器 · 用大白话讲清楚

CutClip.one

作者:leovegas版本 1.0.0i★ –i在云端运行
注册名称one.cutclip/cutclip

安装

概览

作者提供的说明(英文原文)

AI video editor your agent can drive: plan scenes, edit a multi-track timeline, and render MP4s.

PopNx MCP 的三大支柱

这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。

它能帮你做什么

我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。

  • 连接方式: 在云端运行
  • 传输方式: Streamable HTTP
  • 分类: 设计、图片与音频
  • 你需要填写:Authorization

安装

选择你使用的 AI 应用,然后复制设置。

安装之前

标为机密的值是你账户的钥匙。千万不要分享给别人,也不要粘贴到聊天里。

你需要

你用的是哪个应用?

在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。

这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。

在终端中运行这条命令

claude mcp add --transport http cutclip https://cutclip.one/api/mcp --header 'Authorization: your-value'

运行前请先填写空着的值。

在 Cursor 中打开

或者自己复制设置

粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json

{
  "mcpServers": {
    "cutclip": {
      "url": "https://cutclip.one/api/mcp",
      "headers": {
        "Authorization": "your-value"
      }
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

在 VS Code 中打开

或者自己复制设置

粘贴到项目中的 .vscode/mcp.json

{
  "servers": {
    "cutclip": {
      "type": "http",
      "url": "https://cutclip.one/api/mcp",
      "headers": {
        "Authorization": "your-value"
      }
    }
  }
}

浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。

这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。

你的选择通过 Cookie 保存在这台设备上,无需账户。

需要你填写的值

  • Authorization 机密 必填 HTTP header
    Bearer sfs_... personal token. Create one at https://www.cutclip.one (Settings → Connect an AI client). Free credits on signup.

这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。

它能做什么

工具名称和描述与服务器在 2026年11月7日 报告的完全一致。这些内容由作者撰写,不是我们写的;我们以纯文本形式显示。

  • project.create

    Create a new film project (§11.1)

  • project.list_for_owner

    List every project a user owns, most recently updated first

  • project.update

    Rename a project or change its description

  • project.delete

    Permanently delete a project and everything in it (tracks, scenes, assets, render history). Cannot be undone.

  • project.get_state

    Fetch a project's full current state: project, tracks, scenes, timeline items, placeholders, scenarios

  • project.get_summary

    Fetch a compact project summary (project settings, tracks, scene list) — prefer this over project.get_state for normal conversation context (§16)

  • project.export

    Export a project's full structure (settings, tracks, scenes, timeline, placeholders, transitions, current prompts) as a portable JSON document, suitable for project.import

  • project.import

    Create a brand-new project from a project.export JSON document. Only relinks media whose asset id still exists and is owned by ownerId; anything else comes in unresolved.

  • track.create

    Create a video or audio track on a project (§11.4)

  • track.update

    Rename, mute, lock, or set the volume (0-1) of a track

  • track.delete

    Delete a track

  • track.reorder

    Move a track to a new position among the project's tracks — video and audio tracks share one ordering, so a track can move in between tracks of either type

  • scene.create

    Create a storytelling scene on a project — distinct from a timeline clip (§7.2, §11.3)

  • scene.update

    Update a scene's title, summary, narrative purpose, planned duration, or location (its shots inherit the location unless they override it)

  • scene.reorder

    Move a scene to a new position among the project's scenes

  • scene.delete

    Delete a scene

  • timeline.add_placeholder

    Add a video/audio placeholder clip to a track, ready for a generation prompt (§7.5, §11.5)

  • timeline.move_item

    Move a timeline item to a new start time, and optionally to a different track (must be the same track type — video to video, audio to audio — and the item must not be a transition or have one attached)

  • timeline.duplicate_item

    Duplicate a timeline clip (media or placeholder, not a transition), by default onto the same track immediately after itself, pushing whatever else is later on that track to the right to make room (only fails if the target spot would land inside another clip that starts before it)

  • timeline.trim_item

    Trim a timeline item's start, duration, or source in/out points

  • timeline.set_speed

    Change a clip's playback speed (0.2 = 5x slower, 1 = normal, 1.8 = 1.8x faster). Recomputes the clip's timeline duration for the new speed and ripples everything later on the same track to stay contiguous — the source range consumed is unchanged, only how long it takes to play.

  • timeline.set_volume

    Set a clip's own audio gain (0 = silent, 1 = full). Multiplies with its track's volume — both apply together — so a clip can be ducked or muted independent of the rest of its track.

  • timeline.set_transform

    Place a clip on a video track as a layer: its position, size and opacity in the frame. Video tracks stack by order (higher order is drawn on top), and anything outside a scaled-down or transparent clip shows the tracks beneath — so a logo, watermark or picture-in-picture is an im

  • timeline.split_item

    Split a timeline item into two at an absolute timeline position

  • timeline.delete_item

    Delete a timeline item (and its placeholder, if any)

  • timeline.close_gap

    Ripple a timeline item (and everything after it on the same track) earlier to eat the empty space immediately before it. Fails if there is no gap before the item.

  • timeline.batch

    Apply SEVERAL timeline edits in ONE call — one version check, one result, one undo step, all-or-nothing (if any op fails nothing is applied and the error names the failing op). ALWAYS prefer this over repeated single-edit calls whenever you have two or more timeline edits to make

  • timeline.close_gaps

    Close the empty gaps between SEVERAL clips at once (a ripple: each clip after the first is pulled left against its predecessor, and everything later on the track moves with it, so clips after the span keep their spacing). Select the clips with exactly one of itemIds (every gap in

  • timeline.add_transition

    Add a transition (§4.1: crossfade, smooth_blend, fade_from_black, or fade_to_black — a hard cut needs no call at all, it's just the default). crossfade and smooth_blend require fromItemId and toItemId to be two contiguous, back-to-back clips on the same track, and shift toItemId

  • timeline.add_transitions

    Add the SAME transition to several clips in ONE call (and one undo step) — prefer this over repeated timeline.add_transition when joining a run of clips. Select the clips with exactly one of itemIds, trackId (every clip on a track) or all:true (every clip on every track; trackTyp

  • timeline.remove_transitions

    Remove transitions from several clips in ONE call (and one undo step): every transition whose clip(s) are all selected (itemIds, trackId or all:true, like timeline.add_transitions) — a crossfade/smooth_blend needs both its clips selected, a fade just its own clip. Crossfaded clip

  • timeline.remove_transition

    Remove a transition. For a crossfade or smooth_blend this also shifts the incoming clip back to its pre-transition position.

  • timeline.update_transition_duration

    Change how long an existing fade, crossfade or smooth_blend lasts, in ms. For a crossfade or smooth_blend this re-shifts the incoming clip to keep it anchored to the outgoing clip's end.

  • asset.replace_placeholder

    Point a placeholder at a real uploaded/generated asset (§11.7, §18.2). Preserves the placeholder and its prompt for provenance; sets its status to 'generated'.

  • asset.get

    Fetch one asset's metadata (type, source, mime type, duration/dimensions if known) plus a downloadUrl for its file: a plain HTTP GET, no auth header needed, valid 24h (call again for a fresh one)

  • asset.list

    List every asset uploaded/generated for a project (§11.7)

  • scenario.create

    Stage a concept + treatment + scene list before materializing it onto the timeline (§9, §11.2). Does not touch tracks/scenes/timeline — call scenario.apply_to_timeline next.

  • scenario.get

    Fetch a staged scenario's concept, treatment, and scene list

  • scenario.update

    Revise a staged scenario's concept, treatment, and/or scene list before applying it — e.g. after the user asks for a pacing or tone change. Only the fields provided are changed.

  • scenario.delete

    Delete a staged scenario draft (does not affect anything already applied to the timeline)

  • scenario.apply_to_timeline

    The high-level film-building tool (§34): materializes a staged scenario into real tracks, scenes, placeholders, and timeline items in one atomic, undoable step. Creates a video track with one placeholder per scene; optionally a Music track with one placeholder spanning the whole

  • storyline.get

    Fetch the Scenario Maker draft storyline for a project: the premise (and how many reference images it has), the scenes the user has accepted so far (path), and the current batch of suggested next scenes (candidates).

  • storyline.set_premise

    Set the Scenario Maker premise (initial scene or short scenario), optionally with up to 3 reference images (base64) that define the look and setting. Resets the accepted path and candidates. Needs a premise or at least one image.

  • storyline.set_candidates

    Replace the current batch of suggested next scenes with distinct options you propose, so the user can pick one. Each option should follow from the premise and every scene already in the path.

  • storyline.pick

    Accept one of the current candidates as the next scene in the storyline, optionally with edits. Clears the candidates — suggest new ones afterwards.

  • storyline.update_scene

    Edit a scene already accepted into the storyline path

  • storyline.rewind

    Rewind the storyline: keep only the first `keep` accepted scenes (0 clears the path) and discard the rest and the current candidates, so a different choice can be made at that point.

  • storyline.apply

    Turn the accepted storyline path into real scenes, placeholders, prompts and timeline items in one undoable step (stages it as a scenario and applies it), then clears the draft path. With draftShots (default true) each scene is first broken into a shot list like shot.draft (meter

  • prompt.attach

    Attach a structured prompt document to a scene slot (visual/sound/music/voice) or a placeholder directly (§8, §11.6)

  • prompt.update

    Revise a prompt — creates a new revision rather than overwriting the old one (§17)

  • prompt.get

    Fetch one prompt document by id

  • prompt.list_for_scene

    List every prompt ever attached to a scene, across all slots and revisions

  • prompt.list_revisions

    Get the full revision history (oldest first) for a prompt's slot, given any promptId in that lineage (§17)

  • project.undo

    Undo the most recent operation applied to a project (§14)

  • render.preview

    Start a fast, half-resolution preview export of the project's current timeline to MP4 (§11.8). Takes the same aspect/fit as render.final so the framing can be checked cheaply. Costs 1 credit, charged to the project owner; fails with INSUFFICIENT_CREDITS if they can't cover it.

  • render.final

    Start a full-resolution final export of the project's current timeline to MP4 (§11.8). Costs 5 credits at 1080p (the default), 10 at 2k (2560x1440) or 15 at 4k, charged to the project owner; fails with INSUFFICIENT_CREDITS if they can't cover it. The resolution is the short side

  • credits.get_balance

    Get the credit balance of the project's owner, and what a render costs, in milli-credits (1000 = 1 credit). Check this before starting a render if the balance may be low.

  • credits.list_history

    List the project owner's credit history, newest first: purchases, assistant usage, renders and AI generations, each with its signed amount in milli-credits (1000 = 1 credit; negative = spent), balance after, detail (model, resolution, tokens…) and the project it was for. Also ret

  • render.status

    Check the status/progress of a render job

  • render.get_result

    Get a completed render job's result: the output assetId and a downloadUrl for the rendered file — a plain HTTP GET, no auth header needed, valid 24h (call again for a fresh one)

基础检查i

我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月8日

  • ✓ 通过

    作者与仓库一致

    注册名称已通过域名 cutclip.one 验证(注册中心会核实域名所有权)

  • … 尚未检查

    仍在维护

    我们还没有从 GitHub 读取这个仓库

  • … 尚未检查

    有许可证

    我们还没有从 GitHub 读取这个仓库

  • ✓ 通过

    端点有响应

    在 309 毫秒内响应(HTTP 200)

  • – 不适用

    使用 OAuth 登录

    无需登录

  • ✓ 通过

    工具描述中没有隐藏指令

    已读取 60 个工具描述,未发现针对 AI 的隐藏指令

与“设计、图片与音频”中的其他服务器比较i

  • 星标数多于此分类中其他 2801 个服务器的 0%
  • 发布到注册中心的时间比此分类中其他 2801 个服务器的 92% 更新
  • 基础检查结果等于或优于此分类中其他 2801 个服务器的 100%

与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。

常见问题

CutClip.one 安全吗?

它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。

它需要哪些密钥或代码?

你需要填写:Authorization

它在我的电脑上运行还是在云端?

它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。

它还在维护吗?

GitHub 仓库最后更新于–,最新版本(1.0.0)于3天前进入注册中心。

哪些应用可以使用它?

支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。

有什么可以替代它?

同一分类中有:Figma Context MCP, Claude Real Video, Figma MCP Server

替代选择

同一分类中其他作者的服务器,按星标数排序。

如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i

Figma Context MCP

Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.

作者:GLips广泛使用↓ 16.1万/周★ 1.6万5个月前更新

在你的电脑上运行

Claude Real Video

Let any LLM watch a video locally — and search everything it has ever watched.

作者:HUANGCHIHHUNGLeo广泛使用↓ 1106/周★ 22842个月前更新

在你的电脑上运行

Figma MCP Server

The Figma MCP server brings Figma design context directly into your AI workflow.

作者:figma广泛使用★ 215213个月前更新

在云端运行✓ mcp.figma.com

作者的 README

打开 README

按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。

CutClip.one MCP Server

An AI video editor that your coding agent can drive end to end.

Connect Claude Code, Codex, Cursor, Windsurf or any MCP client to CutClip.one. Your agent can then plan a film, cut it on a multi-track timeline and render the MP4. Every change goes through the same tools as the editor itself, so you can open the project in the browser, see what the agent did, and undo any step.

*A quick tour of www.cutclip.one: copy-ready MCP setup for Claude Code, Codex and Cursor, the editor, and the feature list.*

  • 🌐 Website: https://www.cutclip.one
  • 🔌 MCP endpoint: https://cutclip.one/api/mcp (streamable HTTP, hosted, nothing to install)
  • 📇 Official MCP Registry: one.cutclip/cutclip

What your agent can do

AreaCapabilities
ProjectsCreate, list, update, export and import projects; read the full state or a compact summary
Storyline & scenariosTurn one idea into a premise, pick from suggested story beats, write a scenario and lay it out on the timeline
TimelineAdd, move, trim, split, duplicate and delete clips; speed, volume, position/size/opacity; transitions; close gaps; batch edits
Tracks & scenesCreate, reorder and update video/audio tracks and scenes
AssetsUpload media, swap placeholders for real footage, capture still frames
PromptsDraft and revise image/video prompts for each scene, with full revision history
Cinematic film modeShots with a camera rig, characters and locations with reference images, film look, keyframe and video generation, takes
Music & voiceMusic generation for music placeholders, voice-over clips, lip-sync and re-voicing for shots
RenderingPreview and final renders (up to 4K, several aspect ratios, H.264/H.265), status polling and a signed download link
History & creditsUndo any change; check your credit balance and usage history

111 tools in total. One personal token covers all of them, and it only ever sees your own projects.

Quick start

  1. Sign up at https://www.cutclip.one. New accounts get free credits.
  2. Open Settings → Connect an AI client (MCP) and create a token (it starts with sfs_).
  3. Connect your client:

Claude Code

claude mcp add --transport http cutclip https://cutclip.one/api/mcp \

--header "Authorization: Bearer sfs_YOUR_TOKEN"

Codex (~/.codex/config.toml)

[mcp_servers.cutclip]

url = "https://cutclip.one/api/mcp" http_headers = { Authorization = "Bearer sfs_YOUR_TOKEN" }

Cursor / Windsurf / other clients (mcp.json)

{

"mcpServers": { "cutclip": { "type": "http", "url": "https://cutclip.one/api/mcp", "headers": { "Authorization": "Bearer sfs_YOUR_TOKEN" } } } }

  1. Ask your agent something like:

*"Use cutclip to make a 30-second vertical teaser about a lighthouse keeper: plan five scenes, put placeholders

on the timeline with crossfades, add a music track, and render a preview."*

Example tasks

  • Turn a one-line idea into a storyline, a scenario and a rough cut with placeholders
  • Upload your own footage and have the agent trim, order and crossfade it
  • Re-cut one project for 16:9, 9:16 and 1:1, and render each one
  • Batch edits: "make every clip 10% faster and close all the gaps"
  • Overlays: picture-in-picture, logos and GIFs with position, size and opacity
  • Script your pipeline: generate shots → assemble → render → download, all from your agent

Pricing

Pay-as-you-go credits with no subscription. Editing and reading projects is free. Renders and AI generation cost credits, and every new account gets free credits. See https://www.cutclip.one/pricing.

Support

Report bugs and request features in Issues, or email timofeevld@mail.com.


This repository holds the public listing for the hosted CutClip.one MCP server. The service itself is closed source.

问 PopNx