long-video-mcp
Supports packaging long-form videos for YouTube by generating YouTube-style chapter timestamps (e.g. 0:00 ...) from saved project segments.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@long-video-mcp create a 25-minute video project about the history of coffee"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
long-video-mcp
An MCP (Model Context Protocol) server that helps Claude plan, script, pace and package long-form videos (10 to 60+ minutes).
Tools
Tool | What it does |
| Splits a target duration into Hook, Intro, Chapters and Outro, with advice for each part (saves nothing) |
| Creates and saves a project with an auto-generated structure |
| Lists saved video projects |
| Gives a segment a real title (used for YouTube chapters) |
| Saves the narration of a segment and compares its speaking time with the planned time |
| Analyzes any text: word count, duration, and long sentences or paragraphs |
| Progress report: scripted segments and total runtime versus the target |
| Generates YouTube chapter timestamps ( |
| Splits a script into visual beats, with a new visual every N seconds |
| Checklist for pre-production, recording, editing and publishing |
Resources: project://<project-name> returns the full project data (JSON).
Prompts: write_chapter_script is a template for writing a segment within its time budget.
Related MCP server: youtube-content
Install
Requires Node.js (LTS) and Claude Desktop.
git clone https://github.com/Ehmad9/long-video-mcp.git
cd long-video-mcp
npm install
npm run build # produces build/index.js
npm test # starts the server and calls every toolConnect to Claude Desktop
Open Settings → Developer → Edit Config and add the server, using the absolute path to
build/index.js:
{
"mcpServers": {
"long-video": {
"command": "node",
"args": ["/absolute/path/to/long-video-mcp/build/index.js"]
}
}
}On Windows use forward slashes, for example C:/long-video-mcp/build/index.js.
Then fully quit Claude (from the system tray) and open it again.
Projects are stored in ~/.long-video-mcp by default. Set the VIDEO_MCP_DATA
environment variable to change the location.
Example prompts
"Create a 25-minute video project about the history of coffee"
"Write the script for chapter 3, save it, and tell me if it is too long"
"Generate YouTube chapters for the project"
"Make a shot list for this script"
Notes
Never use
console.loginside the server, because it corrupts the stdio channel. Useconsole.error.The default narration speed is 150 words per minute. For slower speech, set
wpmto 120-140.
Contributing and security
Pull Requests are welcome, but every change is reviewed by hand before merging, because this server runs on users' computers. See CONTRIBUTING.md.
License
MIT
Available Tools
10 toolsanalyze_scriptB
Analyze any narration text: word count, estimated duration, and pacing warnings (very long sentences or paragraphs).
Inputs: text (required), wpm.
create_projectA
Create and save a long-video project with an auto-generated structure. Use the returned project name in other tools.
Inputs: title (required), topic (required), minutes (required), wpm, +2 more.
generate_timestampsA
Generate YouTube chapter timestamps for a project. Uses actual script lengths when available, otherwise planned times.
Inputs: project (required), use_script_lengths.
list_projectsB
List all saved video projects.
Inputs: no named parameters.
plan_structureA
Plan the segment structure (hook, intro, chapters, outro) of a long-form video for a target duration. Returns time ranges and retention advice. Does not save anything.
Inputs: minutes (required), chapters.
production_checklistA
Return a stage-by-stage production checklist for a long-form video (pre-production, recording, editing, publishing).
Inputs: stage.
project_reportB
Report progress of a project: which segments have scripts, estimated total runtime versus the target.
Inputs: project (required).
rename_segmentB
Give a segment (e.g. 'Chapter 2') a real title. Titles are used for YouTube chapter timestamps.
Inputs: order (required), title (required), project (required).
save_scriptA
Save (or replace) the narration script of one segment and report its estimated speaking time versus the planned time.
Inputs: order (required), script (required), project (required).
shot_listB
Split a narration script into visual beats (a new visual every N seconds) with timecodes, so a long video never stays on one static shot.
Inputs: script (required), wpm, seconds_per_visual.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Most tools target clearly distinct actions: planning (plan_structure), persistence (create_project, list_projects), scripting (save_script, analyze_script), and reporting (project_report, generate_timestamps). The main overlap is plan_structure versus create_project, both producing segment structures, distinguished only by whether they save; an agent could plausibly reach for the wrong one. rename_segment and save_script are close in that both edit segment content, but their descriptions separate them adequately.
The dominant pattern is verb_noun snake_case (plan_structure, list_projects, create_project, rename_segment, save_script, analyze_script, generate_timestamps). However, three tools break it with noun-only names: project_report, shot_list, and production_checklist. The set is still readable and mostly predictable, but the mixed convention is noticeable.
Ten tools is a well-scoped size for a long-form video planning server. Each tool maps to a distinct stage of the workflow—planning, project persistence, scripting, analysis, reporting, and production guidance—and none feels redundant filler.
The surface covers the core lifecycle: plan a structure, persist a project, title segments, write and analyze scripts, report progress, generate timestamps, and build shot lists and production checklists. Gaps are modest—there is no delete_project or project rename, and no way to remove or list individual segments explicitly, but these are workable around the create/list/report tools.
Related MCP Connectors
Create long-form YouTube videos end to end: script, storyboard, voiceover, final MP4.
The agentic video editor for ChatGPT and Claude. Ask in your chat and it cuts, captions, adds music, titles and motion graphics on a real timeline you can open and adjust.
Plan, render, fix and publish multi-scene AI videos: scripts, storyboards, cast, narration, YouTube.
The YouTube autopilot: plan, write and schedule YouTube videos, and upload from your machine.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables Claude to generate AI videos using ViewMax Studio's video generation tools, including prompt and script creation, smart model selection, cost preview, and task status tracking.2-
- FlicenseNot gradedqualityDmaintenanceEnables AI-powered YouTube content management using Claude Code to orchestrate research, scripting, and scheduling via Google Sheets, Docs, and Drive.-
- FlicenseNot gradedqualityAmaintenanceEnables Claude to read, edit, and generate real Final Cut Pro timelines from exported project XML, handling cut detection, markers, roles, silence removal, rough cuts, keyframes, speed ramps, and transcript-based editing. Every write is sha256-journaled and reversible, with quote-then-confirm generation and optional pushes straight into a running Final Cut Pro.-
- AlicenseNot gradedqualityCmaintenanceEnables Claude to create, edit, validate, and animate video projects through structured project files, including rendering frames or exports, inspecting previews, and applying themes, templates, effects, and subtitles.MIT