# samuelgursky/davinci-resolve-mcp [Health: Active]

**Category:** 🎥 Multimedia Process  
**Repository:** https://github.com/samuelgursky/davinci-resolve-mcp  
**GitHub Stars:** 3830  
**npm Downloads (last month):** 21491  
**Views:** 12  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/samuelgursky-davinci-resolve-mcp

## Description
MCP server integration for DaVinci Resolve providing powerful tools for video editing, color grading, media management, and project control

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "davinci-resolve-mcp": {
    "command": "npx",
    "args": ["-y","davinci-resolve-mcp"]
  }
}
```

## Documentation

## What samuelgursky/davinci-resolve-mcp does

samuelgursky/davinci-resolve-mcp gives MCP-compatible assistants access to DaVinci Resolve through the application’s official scripting interface. Its scope covers video-editing operations, media-pool organization, render configuration, review markers, color work, Fusion, Fairlight, project lifecycle tasks, extension authoring, and source-safe media analysis.

The package includes a local browser control panel. It can show Resolve state, run analysis that does not modify source media, display analyzed clips and shots, and allow analysis results to be edited in the browser.

## How it works

The default compound server is started from `src/server.py` and groups related Resolve actions behind parameters. This reduces the number of tools an assistant needs to load into context. A full mode is available when separate tools for individual Resolve API methods are preferable; it exposes 390 granular operations through `src/server.py --full` or `src/resolve_mcp_server.py`.

The server normally communicates with DaVinci Resolve through external scripting. Resolve Studio must be open, with external scripting set to `Local` under Preferences > General. The package also supports an authenticated loopback bridge for situations where external scripting is unavailable. The bridge is launched from inside Resolve and is selected automatically when needed.

## Setup and configuration

The npm setup command is:

```bash
npx davinci-resolve-mcp setup
```

The installer creates a managed Python environment, detects Resolve installation paths, and can configure MCP clients including Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Zed, Continue, Cline, Roo Code, Codex CLI, and JetBrains IDEs. Source installation is also available by cloning the repository and running `python install.py`.

The project targets Python 3.10 or newer and DaVinci Resolve 18.5 or newer. To use the local control panel from a repository checkout, run:

```bash
venv/bin/python -m src.control_panel
```

The panel listens on loopback and requires the access-token URL printed when it starts. Bridge configuration is stored at `~/.config/davinci-resolve-mcp/bridge.json` by default. Set `DAVINCI_RESOLVE_BRIDGE_CONFIG` to use another location, or set `DAVINCI_RESOLVE_BRIDGE=1` to require bridge transport.

## Tools and capabilities

- Grouped editing and Resolve project actions in compound mode
- One-tool-per-method access through full mode
- Media-pool and project organization
- Render setup and project lifecycle control
- Review markers, color grading, Fusion, and Fairlight operations
- Local clip and shot analysis through the control panel

## Limitations and notes

The free edition cannot normally accept external scripting connections. The in-app bridge provides a documented route for supported free-edition configurations, but Python scripting access changed in Resolve 21.1, so the bridge path should be treated as dependent on the Resolve version. Studio users should use the normal local external-scripting configuration when available.

The bridge uses loopback communication with HMAC-signed requests and one-use nonces. On macOS, Resolve’s Python discovery requires particular interpreter paths or a configured `PYTHON3HOME`; applications launched from the Dock may not inherit shell environment variables. The control panel is local-only and rejects requests without its launch token.

_Full upstream README: https://allmcps.com/mcp/samuelgursky-davinci-resolve-mcp/readme_