DaVinci Resolve MCP Server
Provides 189 tools across 12 domains to control DaVinci Resolve from any MCP-compatible client, including playback, project management, media, timelines, color grading, rendering, Fusion, gallery, and Fairlight.
README
DaVinci Resolve MCP Server
The most complete MCP server for DaVinci Resolve's Scripting API. Control projects, timelines, media, color grading, rendering, and more from any MCP-compatible client.
Features
- 189 tools across 12 domains (playback, project, media storage, media pool, clips, timelines, timeline items, render, color, Fusion, gallery, Fairlight)
- 3 resources for quick context (
resolve://system,resolve://project,resolve://timeline) - Complete API coverage of the DaVinci Resolve Scripting API (Phases 1-3)
- Pydantic v2 models for type-safe inputs and outputs
- Pagination for large clip and timeline item lists
- Auto-reconnect if Resolve restarts or becomes unresponsive
- Platform auto-detection for macOS, Windows, and Linux
Requirements
- DaVinci Resolve 18+ (Free or Studio)
- Python 3.11+
- macOS, Windows, or Linux
Quick Start
Install from PyPI
pip install davinci-resolve-mcp
Or with uv:
uv pip install davinci-resolve-mcp
Claude Desktop Configuration
Add to your Claude Desktop config file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"davinci-resolve": {
"command": "davinci-resolve-mcp"
}
}
}
Then restart Claude Desktop. DaVinci Resolve must be running for tools to work.
See docs/CLAUDE_DESKTOP.md for a detailed setup guide with troubleshooting.
Development Setup
git clone https://github.com/luquimbo/davinci-resolve-mcp.git
cd davinci-resolve-mcp
uv sync --dev
uv run davinci-resolve-mcp
Verify Your Setup
python scripts/check_resolve.py
This prints Resolve version, current project, and timeline info to confirm the scripting bridge is working.
Tool Domains
| Domain | Module | Tools | Description |
|---|---|---|---|
| Playback | playback |
7 | Page navigation, timecode, playhead position, version info |
| Project | project |
21 | CRUD, settings, import/export (.drp), database folders, archive/restore, DB list |
| Media Storage | media_storage |
4 | Browse mounted volumes, list files, import to media pool |
| Media Pool | media_pool |
14 | Folder CRUD, clip management, timeline creation, metadata export |
| Clips | media_pool_item |
21 | Clip metadata, properties, colors, markers, flags, proxy, transcription, replace |
| Timeline | timeline |
29 | CRUD, tracks, items, markers, export (AAF/EDL/FCPXML), compound/Fusion clips, scene detection, auto-subtitles, settings |
| Timeline Items | timeline_item |
27 | Transform, crop, composite, color, markers, flags, takes, unique ID, stabilize, smart reframe, nodes |
| Render | render |
14 | Formats, codecs, presets, job queue, start/stop, progress monitoring |
| Color | color |
24 | Nodes, LUTs, CDL, grade versions, DRX application, color groups, node labels |
| Fusion | fusion |
11 | Compositions CRUD, generators, titles, tool listing |
| Gallery | gallery |
14 | Still albums, grab/import/export stills, PowerGrades, grade application |
| Fairlight | fairlight |
3 | Audio insertion, presets listing, preset application |
Resources (read-only context, no tool call needed):
| URI | Description |
|---|---|
resolve://system |
Resolve version, product name, current page |
resolve://project |
Project name, timeline count, resolution, frame rate |
resolve://timeline |
Timeline name, frame range, track counts, timecode |
Architecture
src/davinci_resolve_mcp/
server.py FastMCP instance, registers all modules, CLI entry point
resolve_api.py Lazy singleton with platform auto-detection and health checks
models.py Pydantic models (CDLValues for ASC color correction)
constants.py Enums for pages, track types, clip colors, marker colors, export types
exceptions.py ResolveNotRunning, ResolveOperationFailed
tools/
playback.py Page navigation, timecode, version
project.py Project CRUD, settings, folders, import/export
media_storage.py Volume browsing, file listing, import
media_pool.py Folder CRUD, clip management, timeline creation
media_pool_item.py Clip metadata, properties, markers, flags
timeline.py Timeline CRUD, tracks, items, markers
timeline_item.py Transform, crop, composite, color, markers
render.py Formats, codecs, presets, jobs, rendering
color.py Nodes, LUTs, CDL, versions, groups
fusion.py Compositions, generators, titles
gallery.py Stills, albums, PowerGrades
fairlight.py Audio insertion, presets
resources/
system_info.py resolve://system
project_info.py resolve://project
timeline_info.py resolve://timeline
Key patterns:
- Lazy singleton (
ResolveAPI): Connects to Resolve on first access, not at import time. AGetVersion()health check runs on every access to detect stale references and reconnect automatically. - One module per domain: Each tool module exports a
register(mcp)function that decorates tool functions with@mcp.tool(). - Defensive error handling: Every API call catches
AttributeError(stale scripting bridge) and re-raises asResolveNotRunning. Version-specific methods are wrapped so missing APIs return clear errors instead of crashes. - Pagination: Large lists (clips, timeline items) support
offsetandlimitparameters, returninghas_moreto signal additional pages.
See docs/ARCHITECTURE.md for the full design document.
Testing
# Unit tests (no Resolve needed)
uv run pytest
# Integration tests (requires Resolve running)
uv run pytest -m integration
Scope
All 189 tools ship in v0.1.0 — covering core editing, color grading, Fusion compositions, gallery stills, Fairlight audio, timeline export, compound clips, transcription, takes management, scene detection, auto-subtitles, stabilization, smart reframe, project archival, and database management.
Auto-Generated Tool Reference
Run the following to generate a complete tool reference from the registered MCP tools:
python scripts/generate_tool_docs.py
This creates docs/TOOLS.md with every tool name, description, parameters, and return type grouped by domain.
Contributing
See CONTRIBUTING.md.
License
MIT. See LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。