resolve-assistant

resolve-assistant

AI-powered video editing assistant that uses Google Gemini to analyze footage, plan professional edits, and build timelines in DaVinci Resolve — automatically. Give it a folder of clips and an editing instruction. It watches every frame, plans the edit, and builds the timeline.

---

What is resolve-assistant?

resolve-assistant is an MCP server that turns raw footage into edited timelines using AI. It combines Google Gemini's multimodal video understanding with DaVinci Resolve's professional editing engine.

The workflow is simple:

1. Point it at a folder of clips → Gemini watches every frame, identifies A-roll vs B-roll, transcribes speech, rates quality, tags camera movements, and analyzes music structure
2. Give it an editing instruction → "Cut a 60-second highlight reel with high energy" or "Build a narrative interview piece, lead with the strongest soundbite"
3. It builds the timeline → Gemini plans every cut, selects the best footage, writes a voiceover script, generates director's notes, and constructs the timeline directly in DaVinci Resolve

No manual logging. No spreadsheets. No bin diving. The AI watches your footage and edits it.

Why resolve-assistant?

• AI watches your footage — Gemini processes every video and audio file with frame-level analysis, not just thumbnails or metadata
• Professional editorial decisions — A-roll/B-roll classification, camera movement detection, speech transcription, take quality scoring, filler word detection, music structure analysis
• Complete edit output — timeline built in Resolve + FCP7 XML backup + director's notes + voiceover script + music production brief
• Works with any MCP client — Claude Desktop, Claude Code, Cursor, or any MCP-compatible AI assistant
• Sidecar architecture — analyze once, edit forever. Sidecar JSONs persist next to your media so re-edits are instant
• One-line install — uvx resolve-assistant or pip install resolve-assistant

---

Quick Start

Prerequisites

1. Google Gemini API key — free at aistudio.google.com
2. DaVinci Resolve (Free or Studio) running with scripting enabled (Preferences → System → General → External scripting using = Network)
3. Python 3.11+
4. ffmpeg — brew install ffmpeg (macOS) or apt install ffmpeg (Linux)

Install

# Using uvx (recommended — runs in isolated environment)
uvx resolve-assistant

# Using pip
pip install resolve-assistant

# Using pipx
pipx install resolve-assistant

Configure Claude Desktop

Add to your Claude Desktop MCP config:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "resolve-assistant": {
      "command": "uvx",
      "args": ["resolve-assistant"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  }
}

Windows note: If uvx is not on your system PATH, use the full path to uvx.exe (e.g., C:\\Users\\YOU\\.local\\bin\\uvx.exe) as the "command" value.

Configure Claude Code

Add to your project's .mcp.json:

{
  "mcpServers": {
    "resolve-assistant": {
      "command": "uvx",
      "args": ["resolve-assistant"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  }
}

---

Using with resolve-mcp — The Complete AI Video Editing Toolkit

resolve-assistant handles AI-driven footage analysis and automatic editing. For direct control over DaVinci Resolve — managing projects, editing timelines, color grading, rendering, Fusion, Fairlight, and 215+ other operations — use resolve-mcp.

They are independent MCP servers that work together. resolve-assistant gives the AI the ability to watch your footage and build edits automatically. resolve-mcp gives the AI hands-on control over every feature in Resolve. Together, they form the most complete AI video editing toolkit available.

Install Both

# Install both with one command
pip install resolve-mcp[assistant]

# Or install separately
pip install resolve-assistant
pip install resolve-mcp

Configure Both Servers Together

Each server needs its own entry in your MCP config. Here is the complete configuration: