🎬 DaVinci Resolve MCP Bridge

The only MCP server for DaVinci Resolve that works on the Free version.

“Add a marker at 5 seconds.” “Transcribe my timeline.” “Remove the background from clip 1.” “Render to MP4.”

Just talk to your AI assistant. It controls Resolve for you.

---

🤔 What Is This?

This project lets AI assistants like Cursor, Claude, or Windsurf control DaVinci Resolve through natural language. Instead of clicking through menus, you just tell the AI what to do — and it does it.

Think of it like giving your AI assistant a pair of hands inside DaVinci Resolve.

---

⚡ What Makes This Different?

Every other DaVinci Resolve MCP server requires the $295 Studio version because they use “external scripting” — a feature Blackmagic locks behind the paywall.

This project works around that entirely.

Instead of calling Resolve from the outside, a small bridge script runs inside Resolve (through the Scripts menu, which is available to everyone). That bridge opens a local connection, and the MCP server talks to it. Simple — and it works on the Free version.

Your AI Assistant (Cursor, Claude, etc.)
         │
         │  talks MCP
         ▼
  resolve_mcp_bridge.py     ← runs on your machine
         │
         │  talks HTTP (localhost)
         ▼
  CursorBridge.py           ← runs INSIDE Resolve (Workspace > Scripts)
         │
         ▼
  DaVinci Resolve API       ← full read + write access

155 of 162 tools work on Free. The 7 that don’t are Studio’s Neural Engine features — and for each one, this project includes a free, local AI replacement that runs on your CPU.

---

🧠 Built-In AI That Replaces $295 Studio Features

No API keys. No cloud. No subscriptions. These run locally on your machine using open-source models:

Models download automatically on first use. Everything runs on CPU — no GPU required.

---

🎛️ 162 Tools Across Every Feature of Resolve

This isn’t a demo with 5 tools. It covers the entire DaVinci Resolve experience:

---

🚀 Getting Started

What You Need

• DaVinci Resolve 18+ (Free or Studio — both work)
• Python 3.9+ on the same machine as Resolve
• Cursor, Claude Desktop, or any MCP-compatible AI assistant

Step 1 — Install the bridge script inside Resolve

Copy src/CursorBridge.py to your Resolve scripts folder:

Step 2 — Set up the MCP server

git clone https://github.com/hiteshK03/davinci-resolve-mcp.git
cd davinci-resolve-mcp
python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate
pip install -r requirements.txt

Step 3 — Tell your AI assistant about it

Add to your .cursor/mcp.json (or equivalent):

{
  "mcpServers": {
    "davinci-resolve": {
      "command": "python",
      "args": ["path/to/davinci-resolve-mcp/src/resolve_mcp_bridge.py"]
    }
  }
}

Step 4 — Start using it

1. Open DaVinci Resolve
2. Go to Workspace → Scripts → CursorBridge
3. Console shows: Bridge is running (read + write)
4. Talk to your AI assistant — it now controls Resolve

---

💬 Things You Can Say

"What's on my timeline right now?"
"Import these files into the media pool"
"Add a green marker at 5 seconds called 'intro ends'"
"Insert a Text+ title at the playhead"
"Set the first clip on track 2 to 70% opacity"
"Zoom in clip 1 to 120% and shift it left"
"Transcribe my timeline audio"
"Isolate the vocals from my timeline"
"Remove the background from clip 1 on video track 1"
"Apply this LUT to node 1 of the current clip"
"Set up an MP4 H.265 render and start it"
"Export the timeline as FCPXML"
"Grab a still from the current frame"

---

🔁 Works With the Full Pipeline

This is one piece of a three-server video production setup:

The AI assistant orchestrates across all three automatically. Generate an image → import into Resolve → place on timeline → adjust properties — all from a single conversation.

---

📋 Free vs Studio Compatibility

155/162 tools work on Free. The 5 that don’t have local AI replacements. Only 2 have no alternative (Smart Reframe, Stabilization).

---

⚠️ Limitations

• Keyframe animations — the scripting API only supports static property values, not animated keyframes
• Fusion node parameters — text content and effect values inside compositions need the Fusion page UI
• Transitions — must be added manually from the Effects Library
• Background removal on video — CPU-bound, can be slow for long clips
• Gallery stills — require being on the Color page

---

📄 License

MIT — use it however you want.