kdenlive-mcp-server
MCP 服务器:Kdenlive
一个 Python MCP 服务器,让 AI 代理能够通过自然语言指令在 Kdenlive 中执行专业视频编辑。
这并不是对少量 Kdenlive 命令的简单封装。它是一个分层系统:
AI Agent (Claude)
-> MCP tools (project/media/timeline/motion/effects/audio/...)
-> Video analysis layer (ffprobe, librosa)
-> Internal timeline model (Project / Sequence / Track / Clip / Keyframe)
-> Kdenlive adapter (translates the model to/from real .kdenlive XML)
-> .kdenlive project file
-> Kdenlive / FFmpeg / MLT (melt) renderingAI 从不直接编辑 .kdenlive XML。它调用类型化的 MCP 工具来修改一个干净、可测试的内部时间线模型;一个专用适配器在保存时将模型转换为真实的 Kdenlive 项目。
状态
正在积极开发中,按阶段构建(见下文 PHASES)。核心时间线模型、Kdenlive XML 适配器、效果/转场目录以及音频/节拍同步引擎均已实现,并已针对真实的 Kdenlive 26.04.3 + MLT 安装进行测试。面向 MCP 工具的、向 AI 代理暴露所有功能的层正在下一步构建中。
Related MCP server: opencut-mcp
为什么基于真实文件格式
适配器(src/kdenlive_mcp/kdenlive/adapter/)所使用的 Kdenlive 项目 XML 结构,是从真实的 .kdenlive 项目中逆向工程而来,并与本地安装的 Kdenlive 26.04.3 + MLT 7.38(melt)交叉验证,而非凭空猜测。效果和转场目录直接解析自 Kdenlive 自带的官方效果/转场 XML 定义,因此 AI 永远不会引用到已安装 Kdenlive 中实际不存在的效果。
项目布局
src/kdenlive_mcp/
core/ # internal model: timeline, keyframes, effects, transitions, audio, assets
kdenlive/ # the adapter: XML read/write, effect & transition catalogs, capability detection
media/ # ffmpeg/ffprobe subprocess wrappers, thumbnail generation
storage/ # workspace path safety, caching, snapshots
validation/ # project validation
mcp_tools/ # MCP-facing tool definitions (in progress)
tests/开发
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
pytest需要 ffmpeg/ffprobe 位于 PATH 中。kdenlive/melt 为可选,但启用后可以解锁效果/转场目录,并针对真实安装进行结构验证。
This server cannot be deployed
Maintenance
Related MCP Connectors
AI video editor for agents and humans: timeline, captions, color, audio and generation as MCP tools.
MCP server for Clipkit — gives AI agents a video toolbox via the Clipkit schema.
Autonomous AI video editor over MCP: viral clips, captions, reframe, chroma key, MP4 export
Agentic AI video editor: edit a real timeline, generate clips, music and voice, then render.
Related MCP Servers
- AlicenseAqualityCmaintenanceAn MCP server that lets AI agents visually perceive and edit video inside DaVinci Resolve Studio by rendering frames and manipulating the timeline.21MIT
- AlicenseNot gradedqualityDmaintenanceMCP server that enables LLM agents to control the OpenCut classic video editor's timeline through a Playwright-controlled browser, allowing automated video editing tasks like adding media, inserting clips, splitting, trimming, and exporting.7MIT
- FlicenseNot gradedqualityBmaintenanceMCP server for DaVinci Resolve video editing, enabling AI agents to execute editing operations, manage creative libraries, and analyze media content.-
- AlicenseNot gradedqualityCmaintenanceA real video editor for AI agents, served over MCP, enabling journaled timeline editing, rendering via FFmpeg/MLT, and deterministic CLI operation.1MIT