{"id":1546,"date":"2026-09-11T08:53:36","date_gmt":"2026-09-11T11:53:36","guid":{"rendered":"https:\/\/www.linuxpro.com.br\/?p=1546"},"modified":"2026-09-11T08:59:48","modified_gmt":"2026-09-11T11:59:48","slug":"openchatcut-editor-de-video-com-ia-local-e-mcp","status":"publish","type":"post","link":"https:\/\/www.linuxpro.com.br\/en\/2026\/09\/openchatcut-editor-de-video-com-ia-local-e-mcp\/","title":{"rendered":"OpenChatCut: local AI video editor with MCP"},"content":{"rendered":"<p><img loading=\"lazy\" decoding=\"async\" src=\"\/wp-content\/uploads\/2026\/09\/openchatcut-editor-video-ia.webp\" alt=\"Mascote LinuxPro editando uma timeline no OpenChatCut, com o cachorro caramelo ao lado\" width=\"1486\" height=\"856\" \/><\/p>\n<p>A video editor with AI is only worth it if the result remains under your control. The <a href=\"https:\/\/openchatcut.com\/\" target=\"_blank\" rel=\"noopener\">OpenChatCut<\/a> proposes exactly that: local projects, an editable multi-track timeline and agents that work on clips, transitions, subtitles and audio \u2014 without turning the video into a black box that's impossible to adjust afterward.<\/p>\n<h2>What is OpenChatCut<\/h2>\n<p>OpenChatCut is a video editor <em>open source<\/em>, distributed under <a href=\"https:\/\/github.com\/0xsline\/OpenChatCut\/blob\/main\/LICENSE\" target=\"_blank\" rel=\"noopener\">AGPL-3.0<\/a>, which combines a professional timeline with AI-powered conversational editing. It is an independent project and has no affiliation with the commercial product ChatCut.<\/p>\n<p>The key difference lies in where the editing takes place. Instead of simply requesting a finished video from a generator, the agent writes changes to an actual project: video and audio tracks, clips, cuts, transitions, effects, subtitles, and media. You can then still review the timeline, undo an action, manually adjust, and export when satisfied.<\/p>\n<ul>\n<li><strong>Timeline:<\/strong> multiple tracks, cuts, clip splitting, ripple editing, markers, keyframes, undo and redo.<\/li>\n<li><strong>Text and subtitles:<\/strong> word-level transcription editing, pause and speaker identification, subtitle styles and SRT export.<\/li>\n<li><strong>Image and sound:<\/strong> transitions, LUTs, WebGL effects, music, sound effects, voice recording, <em>ducking<\/em> and vocal isolation.<\/li>\n<li><strong>Deliverables:<\/strong> MP4 video, audio, subtitles, FCPXML, and full project data.<\/li>\n<\/ul>\n<h2>Local-first doesn't mean AI without network<\/h2>\n<p>By default, projects, media, and history stay on the user's machine. This is useful for anyone who doesn't want to automatically send client material to an online service. Manual editing and local timeline don't depend on a configured AI service.<\/p>\n<p>But it's worth separating local storage from local processing: model requests, image or video generation, online media search, and transcription depend on the providers you configure. Before importing sensitive material, check which integrations will be used, where the keys are stored, and the vendor's policy. The documentation states that the keys stay on the local server side, not in the browser.<\/p>\n<h2>Agents, Skills and MCP on the timeline<\/h2>\n<p>The project includes an integrated agent and also exposes a local MCP endpoint. This enables compatible clients\u2014like Codex and Claude Code\u2014to inspect and prepare changes in an open project. The documented flow creates an isolated editing session, logs operations to draft, and only then sends them for review.<\/p>\n<p>This design is safer than giving the agent unrestricted control of the editing island. In manual mode, review the proposal in OpenChatCut before applying it. Even in automatic mode, work on a copy when the delivery is important: check cuts, synchronization, subtitles, volume, and export files. For an overview of how AI clients use tools in the terminal, also read <a href=\"\/en\/2026\/09\/assistentes-de-ia-no-terminal-linux\/\">AI assistants on Linux<\/a> and <a href=\"\/en\/2026\/09\/vibe-coding-linux-skills-ia-cli\/\">AI skills in the CLI<\/a>.<\/p>\n<p>The MCP endpoint is local by default. If you decide to expose it on the network, treat it as a privileged editing interface: use Bearer token, restrict access to the necessary network, and don't publish the port without authentication.<\/p>\n<h2>Download and install on Windows, macOS and Linux<\/h2>\n<p>The simplest path is to open the <a href=\"https:\/\/github.com\/0xsline\/OpenChatCut\/releases\" target=\"_blank\" rel=\"noopener\">OpenChatCut Releases page<\/a> and download the file for the correct architecture and operating system. It is the reference for the latest version, fixes, and SHA 256 sums.<\/p>\n<table>\n<thead>\n<tr>\n<th>System<\/th>\n<th>Published File<\/th>\n<th>Installation<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>Windows x64<\/strong><\/td>\n<td>Installer <code data-no-translation=\"\">.exe<\/code><\/td>\n<td>Download the installer, run it, and follow the Windows wizard.<\/td>\n<\/tr>\n<tr>\n<td><strong>macOS Apple Silicon<\/strong><\/td>\n<td>Image <code data-no-translation=\"\">arm64.dmg<\/code><\/td>\n<td>Open the DMG and install the application. Current builds for macOS may require manual system approval on first open, as they are not yet signed or notarized.<\/td>\n<\/tr>\n<tr>\n<td><strong>macOS Intel<\/strong><\/td>\n<td>Image <code data-no-translation=\"\">x64.dmg<\/code><\/td>\n<td>Open the DMG and install the application. Choose this package only for Intel Macs.<\/td>\n<\/tr>\n<tr>\n<td><strong>Linux x86_64<\/strong><\/td>\n<td>Package <code data-no-translation=\"\">.AppImage<\/code><\/td>\n<td>Download, make the file executable and run it; a traditional installer is not needed.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>On Linux, go to the download folder and use the name of the AppImage you downloaded:<\/p>\n<pre data-no-translation=\"\"><code class=\"language-bash\" data-no-translation=\"\">chmod +x OpenChatCut-*-x86_64.AppImage\n.\/OpenChatCut-*-x86_64.AppImage<\/code><\/pre>\n<p>Before opening a production project, test media import and a short export. OpenChatCut is in active development; downloading the current release instead of copying old links reduces the risk of getting a replaced version.<\/p>\n<h3>Run from source code<\/h3>\n<p>To test or develop from the repository, the documentation requires Node.js 24.x and npm:<\/p>\n<pre data-no-translation=\"\"><code class=\"language-bash\" data-no-translation=\"\">git clone https:\/\/github.com\/0xsline\/OpenChatCut.git\ncd OpenChatCut\nnpm install\ncp .env.example .env.local\nnpm run dev<\/code><\/pre>\n<p>Then, open <code data-no-translation=\"\">http:\/\/localhost:5199<\/code>. Start without model credentials if the intent is to explore the interface and local editing; configure in <code data-no-translation=\"\">.env.local<\/code> only the services you really need. Never send this file to Git, as it may contain provider keys.<\/p>\n<p>To connect an external agent, the repository proposes installing the skills router:<\/p>\n<pre data-no-translation=\"\"><code class=\"language-bash\" data-no-translation=\"\">npx skills add 0xsline\/OpenChatCut<\/code><\/pre>\n<p>The project documentation describes this installation as the way to register the local MCP connection and load specialized skills when needed. Do this only on a workstation you control and keep OpenChatCut updated before giving an agent access to local media.<\/p>\n<h2>Is OpenChatCut an alternative to ChatCut?<\/h2>\n<p>Yes, in the sense of category: both use AI to help edit video through natural language. But they are not the same product, nor is OpenChatCut an official version or fork of ChatCut. OpenChatCut presents itself as an independent alternative, under AGPL-3.0-or-later, focused on local project and editable timeline.<\/p>\n<table>\n<thead>\n<tr>\n<th>Criterion<\/th>\n<th>ChatCut<\/th>\n<th>OpenChatCut<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>Usage model<\/strong><\/td>\n<td>Commercial hosted editor, accessed via web and on the product account.<\/td>\n<td>Open source application to run locally, with public code and AGPL license.<\/td>\n<\/tr>\n<tr>\n<td><strong>AI workflow<\/strong><\/td>\n<td>The site features prompt-based editing and transcription, captions, image generation, video, motion graphics and music.<\/td>\n<td>The integrated agent or an MCP client modifies a multi-track timeline, maintaining history, undo and manual editing.<\/td>\n<\/tr>\n<tr>\n<td><strong>Data and project<\/strong><\/td>\n<td>Workflow tied to the platform and service account.<\/td>\n<td>Projects and media are local-first by default; AI services are optional integrations.<\/td>\n<\/tr>\n<tr>\n<td><strong>Costs<\/strong><\/td>\n<td>The commercial site works with plans and credits; check prices and limits directly before signing up.<\/td>\n<td>The open-source editor can be downloaded and run without a product license, but models, generation, and other external services may charge separately.<\/td>\n<\/tr>\n<tr>\n<td><strong>When it makes sense<\/strong><\/td>\n<td>For those who prefer a managed service and a ready-made cloud workflow.<\/td>\n<td>For those who prioritize auditable code, local files, timeline control, and MCP integration with Codex or Claude Code.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>The comparison is not a performance test nor a promise of feature-for-feature parity. The <a href=\"https:\/\/chatcut.io\/\" target=\"_blank\" rel=\"noopener\">ChatCut<\/a> has a strong focus on in-product transcription editing and generation; the <a href=\"https:\/\/openchatcut.com\/blog\/chatcut-vs-openchatcut\" target=\"_blank\" rel=\"noopener\">OpenChatCut guide<\/a> makes it clear that its proposal is to prioritize openness, local control, and agent tools. If your work depends on a specific feature, do a pilot project in both before migrating a production deployment.<\/p>\n<p>To learn more, read the project's own text about <a href=\"https:\/\/openchatcut.com\/blog\/open-source-chatcut-alternative\" target=\"_blank\" rel=\"noopener\">why it defines itself as an open-source alternative to ChatCut<\/a> and the detailed comparison <a href=\"https:\/\/openchatcut.com\/blog\/chatcut-vs-openchatcut\" target=\"_blank\" rel=\"noopener\">ChatCut vs OpenChatCut<\/a>.<\/p>\n<h2>Resource library and open format<\/h2>\n<p>A <a href=\"https:\/\/openchatcut.com\/resources\" target=\"_blank\" rel=\"noopener\">community library<\/a> gathers motion graphics animations, sound effects, transitions, visual effects, zooms, LUTs and skills. Each resource brings license and version requirements: read this data before including a pack in commercial or client work.<\/p>\n<p>, with runtime validation. <code data-no-translation=\"\">openchatcut-plugin@1<\/code>, The official library materials are MIT; community contributions may adopt another license. It's a small detail, but avoids discovering restrictions when the video is already ready.<\/p>\n<h2>Is it worth it?<\/h2>\n<p>OpenChatCut makes more sense for those who want to experiment with AI in editing without giving up the timeline, files, and the ability to correct each decision. It's especially interesting for interview cuts, short videos, captions, and agent-assisted automations.<\/p>\n<p>The project is in active development; formats, agent tools and installers may change. Use a test project first, keep copies of the original material and compare the result with alternatives that also cover editing on Linux, such as <a href=\"\/en\/2026\/09\/como-instalar-drift-ubuntu-editor-video-ia\/\">Drift<\/a> and the <a href=\"\/en\/2026\/09\/davinci-resolve-21-1-mcp-agentes-ia\/\">DaVinci Resolve with MCP<\/a>. To understand the technical base of rendering and codecs, it's worth revisiting <a href=\"\/en\/2026\/09\/a-historia-do-ffmpeg\/\">the history of FFmpeg<\/a>.<\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/github.com\/0xsline\/OpenChatCut\" target=\"_blank\" rel=\"noopener\">OpenChatCut on GitHub<\/a><\/li>\n<li><a href=\"https:\/\/openchatcut.com\/\" target=\"_blank\" rel=\"noopener\">OpenChatCut official site<\/a><\/li>\n<li><a href=\"https:\/\/openchatcut.com\/docs\" target=\"_blank\" rel=\"noopener\">Official documentation<\/a><\/li>\n<li><a href=\"https:\/\/openchatcut.com\/resources\" target=\"_blank\" rel=\"noopener\">Resource library<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/0xsline\/OpenChatCut\/releases\" target=\"_blank\" rel=\"noopener\">OpenChatCut \u2014 Releases and installers<\/a><\/li>\n<li><a href=\"https:\/\/openchatcut.com\/blog\/open-source-chatcut-alternative\" target=\"_blank\" rel=\"noopener\">OpenChatCut \u2014 Open-source ChatCut alternative<\/a><\/li>\n<li><a href=\"https:\/\/openchatcut.com\/blog\/chatcut-vs-openchatcut\" target=\"_blank\" rel=\"noopener\">OpenChatCut \u2014 ChatCut vs OpenChatCut<\/a><\/li>\n<li><a href=\"https:\/\/chatcut.io\/\" target=\"_blank\" rel=\"noopener\">ChatCut \u2014 official site<\/a><\/li>\n<\/ul>","protected":false},"excerpt":{"rendered":"<p>Meet OpenChatCut, local-first open source editor that combines multi-track timeline, conversational AI and MCP without giving up editable projects.<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[115,2,111],"tags":[437,116,4,311,344],"class_list":["post-1546","post","type-post","status-publish","format-standard","hentry","category-ia","category-linux","category-opensource","tag-edicao-de-video","tag-ia","tag-linux","tag-mcp","tag-video"],"_links":{"self":[{"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/posts\/1546","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/comments?post=1546"}],"version-history":[{"count":2,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/posts\/1546\/revisions"}],"predecessor-version":[{"id":1549,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/posts\/1546\/revisions\/1549"}],"wp:attachment":[{"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/media?parent=1546"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/categories?post=1546"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.linuxpro.com.br\/en\/wp-json\/wp\/v2\/tags?post=1546"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}