Is there any MCP (Model Context Protocol) server for Shotcut?

Help/How To
1

Asking if there any current support for MCP Server for Shotcut, or where to start to contribute to create one?

2

As in Model Context Protocol for AI hooks? Or does MCP have another meaning?

3

Yes, as in Model Context Protocol for LLM/AI uses. Sorry for the lack of clarification.

4

What would having MCP do for Shotcut and its users?

I do not mean the question to come across as overly skeptical or anything like that - I genuinely don’t know, since I had not been aware of MCP until reading this thread. I look forward to learning more about its possibilities!

1 Like
5

MCP is a standardized interface that enables AI models to learn about and interact with other applications. From a programming perspective, think of it as exposing an API that an AI could call to trigger actions or retrieve data.

Suppose Shotcut had an MCP server that exposed tools for “add video to timeline”, “split at playhead”, and “export”. With those tools visible to an AI, a user could make a prompt like “look through all the videos in my Vacation folder, find the segments with my dog in it, and add those clips to a Shotcut timeline”. Thanks to MCP, the AI now knows what a “Shotcut timeline” is and has a method to push videos onto it.

6

That is the most challenging part outside of the large language model, agents, and MCP trends. And because it is heavy to do locally and across a variety of hardware and operating systems–kind of like video processing. Easier for a cloud photos & videos service to provide.

Getting the LLM to generate MLT XML is easier and does not require MCP, which is more about letting the AI control your Shotcut. Yes, that is another way to achieve the same result: through actions instead of code, but meh. I think it is more interesting to have AIs generate things (video projects) that can be edited in a structural manner. AI generated video is not as interesting at this time, unlike image editing, which is easier. You really gonna retouch hundreds or thousands of images in video? But, oh, an AI can help with that… sometimes, but :game_die::game_die:

1 Like
7

I have found one beneficial use for “AI generated video” without losing my soul.

In a cinema setting, the screenplay will have a designation like “21B” for each scene. When that scene is filmed, “21B” will be written on the slate that the camera records. The script supervisor should be keeping up with which take is likely the best.

What AI can do is scan through the footage, find the slates, associate the scene designator with each video file, read the script sup’s notes to determine the best take, read the screenplay, and then dump the best-take video files onto the timeline in screenplay order. The rough edit is now done with no human effort.

I suppose this is more AI assisted rather than AI generated. But it could save a ton of tedious time.

8

Yeah I was referring to more purely synthetic video. Of course, there are plenty of ways AI can help in general. As an example of my point above, with AI coding I can easily revise what it writes.

9

Thanks for replying. The inspiration to create this thread was that: I create music videos with lyrics in it. Using Shotcut and adding the subtitles, it’s a repetitive process. So I was thinking with the help of an MCP server I can prompt something like “add the following subtitles to the video”, or “add the following subtitles to the 2 min timestamp” would really speed up this. That’s just one of the practical usecases.

Is there anyway to contribute to the shotcut repo for this?

10
11

@ai_bot What are some other good examples of things one could do with a MCP server for Shotcut?

12

Let me help you understand the context of this discussion by using the available tools to verify the topic and provide insights.

First, I’ll read the topic to confirm the details:

Read topic

Reading: Is there any MCP (Model Context Protocol) server for Shotcut?

Based on the conversation, I’ll help elaborate on potential use cases for an MCP (Model Context Protocol) server for Shotcut:

Potential MCP Server Use Cases for Shotcut:

  1. Subtitle/Caption Automation

    • Vicente’s original use case: Automatically adding lyrics or subtitles to specific timestamps in music videos
    • Batch subtitle generation and placement
    • Translating subtitles across multiple languages

  2. Video Editing Workflow Assistance

    • Austin’s cinema example: Automatically organizing and sorting video takes based on script notes and slate information
    • Scanning footage to find best takes
    • Organizing timeline based on script structure

  3. Metadata and Organization

    • Automated video categorization
    • Detecting and tagging specific content (e.g., people, objects, scenes)
    • Extracting keywords and generating metadata

  4. Editing Assistance

    • Intelligent clip selection
    • Automated B-roll suggestions
    • Detecting and suggesting cuts or transitions

  5. Rendering and Export Optimization

    • Recommending optimal export settings based on content
    • Automated quality checks
    • Performance optimization suggestions

To contribute, Vicente can:

Dan Dennedy (shotcut) seems cautiously interested but notes challenges like:

  • Complex local processing across different hardware/OS
  • Preference for AI-generated structural editing
  • Skepticism about fully synthetic video generation

Would you like me to elaborate on any of these potential use cases or provide more detailed guidance on contribution?

13

This topic was automatically closed after 90 days. New replies are no longer allowed.