ShotGrid MCP Server

A Model Context Protocol (MCP) server that provides AI assistants with seamless access to Autodesk ShotGrid (Flow Production Tracking)

English | 简体中文

📖 Documentation | 中文文档

Overview

ShotGrid MCP Server enables AI assistants like Claude, Cursor, and VS Code Copilot to interact directly with your ShotGrid (Flow Production Tracking) data. Built on FastMCP, it provides a high-performance bridge between AI tools and production tracking workflows.

Demo

0. Configure ShotGrid MCP in Code Editor

1. Query Task Schedule & Workload Visualization

Prompt: Query the team's task schedule for the past week, calculate workload rate based on 8 hours per day, and visualize it in web format

2. Batch Create Assets & Assign Tasks

Prompt: Batch create the recommended hero characters in the shotgrid Demo:Animation project, categorize them as characters, use the FilmVFX-CharacterAsset task template, assign tasks to Yang Zhuo, with start and end dates set to next week

3. TimeLog Statistics & Visualization

Prompt: Query timelog data from shotgrid and visualize it in web format

4. Department Efficiency Statistics & Send to WeCom

Prompt: Calculate department efficiency and send the data to WeCom. Efficiency formula: Efficiency = Task bid / Timelog hours

More Examples

Features

Quick Start

Installation

# Using uv (recommended)
uv pip install shotgrid-mcp-server

# Or using pip
pip install shotgrid-mcp-server

Configuration

Set your ShotGrid credentials:

export SHOTGRID_URL="https://your-site.shotgunstudio.com"
export SHOTGRID_SCRIPT_NAME="your_script_name"
export SHOTGRID_SCRIPT_KEY="your_script_key"

Usage

stdio Transport (Default) - For Claude Desktop, Cursor, etc.

uvx shotgrid-mcp-server

HTTP Transport - For Remote Access

uvx shotgrid-mcp-server http --host 0.0.0.0 --port 8000

MCP Client Configuration

Add the server to your MCP client configuration:

Claude Desktop

{
  "mcpServers": {
    "shotgrid": {
      "command": "uvx",
      "args": ["shotgrid-mcp-server"],
      "env": {
        "SHOTGRID_URL": "https://your-site.shotgunstudio.com",
        "SHOTGRID_SCRIPT_NAME": "your_script_name",
        "SHOTGRID_SCRIPT_KEY": "your_script_key"
      }
    }
  }
}

Cursor / VS Code / Other MCP Clients

{
  "mcpServers": {
    "shotgrid": {
      "command": "uvx",
      "args": ["shotgrid-mcp-server"],
      "env": {
        "SHOTGRID_URL": "https://your-site.shotgunstudio.com",
        "SHOTGRID_SCRIPT_NAME": "your_script_name",
        "SHOTGRID_SCRIPT_KEY": "your_script_key"
      }
    }
  }
}

HTTP Transport (Remote)

{
  "mcpServers": {
    "shotgrid": {
      "type": "http",
      "url": "http://your-server:8000/mcp"
    }
  }
}

Deployment

See the Deployment Guide for detailed instructions.

Available Tools

This server provides 40+ tools for interacting with ShotGrid:

MCP Apps

shotgrid_dashboard is an MCP App: it returns data and an interactive UI. The tool advertises ui://shotgrid/dashboard in its _meta.ui.resourceUri, and the host renders that resource in a sandboxed iframe.

Hosts without MCP Apps support automatically receive the same summary as plain text.

The UI is bundled into a single self-contained HTML file, so no Node runtime is needed in production. See the MCP Apps guide for the wire format and rebuild instructions.

python scripts/verify_mcp_app.py

Example Prompts

Once connected, you can ask your AI assistant:

• "Find all shots updated last week in Project X"
• "Create a playlist with yesterday's lighting renders"
• "Add a note to SHOT_010 about the background lighting"
• "Summarize time logs for the Animation department this month"
• "Show me a dashboard of Task statuses for Project X"

Development

# Clone and install
git clone https://github.com/loonghao/shotgrid-mcp-server.git
cd shotgrid-mcp-server
pip install -r requirements-dev.txt

# Run tests
nox -s tests

# Development server with hot reload
uv run fastmcp dev src/shotgrid_mcp_server/server.py:mcp

# Verify the Skills extension against a live HTTP server
uv run python scripts/verify_skills_extension.py

Documentation

See the /docs directory for detailed documentation.

Contributing

Contributions welcome! Please follow the Google Python Style Guide, write tests, and use Conventional Commits — commit messages drive versioning and the changelog. See RELEASE.md for how releases are cut.

License

MIT

Architecture

flowchart TB
    subgraph Clients["🤖 MCP Clients"]
        direction LR
        CLAUDE["Claude Desktop"]
        CURSOR["Cursor"]
        VSCODE["VS Code"]
        AI["Other AI"]
    end

    subgraph MCP["⚡ ShotGrid MCP Server"]
        direction LR
        TOOLS["40+ Tools"]
        POOL["Connection Pool"]
        SCHEMA["Schema Cache"]
    end

    subgraph ShotGrid["🎬 ShotGrid API"]
        direction LR
        P["Projects"]
        S["Shots"]
        A["Assets"]
        T["Tasks"]
        N["Notes"]
    end

    Clients -->|"MCP Protocol<br/>stdio / http"| MCP
    MCP -->|"REST API"| ShotGrid

    style Clients fill:#2ecc71,stroke:#27ae60,color:#fff
    style MCP fill:#3498db,stroke:#2980b9,color:#fff
    style ShotGrid fill:#e74c3c,stroke:#c0392b,color:#fff