install
Install
Install cutmaster-ai from PyPI and wire it into Claude Desktop, Claude Code, Cursor, or DaVinci Resolve. Most paths take under 5 minutes.
Before you start
- 01 DaVinci Resolve Studio 18.5+ The $295 paid version. Free edition has no Scripting API. Verify via the About menu — must say "DaVinci Resolve Studio".
- 02 External Scripting enabled Preferences → System → General → External scripting using →
Local. Save and restart Resolve. - 03 Python 3.11 or 3.12 3.13+ has ABI issues with Resolve's
fusionscriptmodule. Earlier versions also unsupported. - 04 uv — required, not optional The committed
.mcp.jsoninvokes the server viauv run; MCP clients won't connect without it. Install withbrew install uvon macOS.
Install the package
$ pip install cutmaster-ai Or with uv:
$ uv add cutmaster-ai Wire it into an MCP client
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"cutmaster-ai": {
"command": "/path/to/.venv/bin/python3",
"args": ["-m", "cutmaster_ai"]
}
}
} Restart Claude Desktop. The cutmaster-ai server should appear in the MCP server list.
The repo ships a .mcp.json file — Claude Code auto-detects it. From a clone:
$ git clone https://github.com/CelaviiHQ/cutmaster-ai.git $ cd cutmaster-ai $ claude
For global access (from any directory), add the MCP server to ~/.claude/settings.json. See the full setup guide ↗.
Any client that speaks MCP stdio works. Configure the command as python -m cutmaster_ai from inside an environment that has cutmaster-ai installed.
Specifics for ten clients live in the universal installer ↗.
One-command setup (macOS / Linux)
$ git clone https://github.com/CelaviiHQ/cutmaster-ai.git $ cd cutmaster-ai $ bash scripts/setup.sh
The script creates a venv, installs dependencies, configures Claude Desktop, and verifies the Resolve connection. Add --all to also install global skills, agents, hooks, and the macOS LaunchAgent auto-start.
Optional: AI features
The Gemini-powered tools (frame analysis, OCR, color assist, editorial critique) require an API key:
$ cp .env.example .env # Edit .env and set: GEMINI_API_KEY=your-key-here
Get a key at aistudio.google.com/apikey ↗. Free tier is enough for typical usage.
Verify the connection
$ python -m cutmaster_ai # Expect: stdio MCP server running. Ctrl-C to stop.
Then in your MCP client, run any read-only tool — e.g. cutmaster_get_project_info — to confirm the bridge is live.
Auto-start on macOS
$ bash scripts/setup.sh --launchagent Installs ~/Library/LaunchAgents/ai.cutmaster.mcp.plist. The MCP server starts on login.
Other clients and the panel
For the React workflow panel (browser today; Resolve dock at v3-6) see the panel setup guide ↗. For the full setup reference covering all ten supported clients, the Workflow Integration plugin, the Claude Code plugin, and packaging, see docs/SETUP.md ↗.