<div align="center">

<img src="images/logo.png" alt="ShotGrid MCP Server Logo" width="200">

# ShotGrid MCP Server

**A [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server that provides AI assistants with seamless access to Autodesk ShotGrid (Flow Production Tracking)**

English | [简体中文](README_zh.md)

[![Python Version](https://img.shields.io/pypi/pyversions/shotgrid-mcp-server.svg)](https://pypi.org/project/shotgrid-mcp-server/)
[![PyPI version](https://badge.fury.io/py/shotgrid-mcp-server.svg)](https://badge.fury.io/py/shotgrid-mcp-server)
[![License](https://img.shields.io/github/license/loonghao/shotgrid-mcp-server.svg)](LICENSE)
[![codecov](https://codecov.io/gh/loonghao/shotgrid-mcp-server/branch/main/graph/badge.svg)](https://codecov.io/gh/loonghao/shotgrid-mcp-server)
[![Downloads](https://static.pepy.tech/badge/shotgrid-mcp-server)](https://pepy.tech/project/shotgrid-mcp-server)
[![Downloads](https://static.pepy.tech/badge/shotgrid-mcp-server/week)](https://pepy.tech/project/shotgrid-mcp-server)
[![Downloads](https://static.pepy.tech/badge/shotgrid-mcp-server/month)](https://pepy.tech/project/shotgrid-mcp-server)

**📖 [Documentation](https://loonghao.github.io/shotgrid-mcp-server/) | [中文文档](https://loonghao.github.io/shotgrid-mcp-server/zh/)**

</div>

## 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](https://github.com/jlowin/fastmcp), it provides a high-performance bridge between AI tools and production tracking workflows.

### Demo

#### 0. Configure ShotGrid MCP in Code Editor

![Configure ShotGrid MCP in Code Editor](images/00-codebuddy-config-shotgrid-mcp.gif)

#### 1. Query Task Schedule & Workload Visualization

![Query Task Schedule & Workload Visualization](images/01-query-projects-visualize-tasks.gif)

**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

![Batch Create Assets & Assign Tasks](images/02-batch-create-assets-tasks-assign.gif)

**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

![TimeLog Statistics & Visualization](images/03-timelog-statistics-visualize.gif)

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

#### 4. Department Efficiency Statistics & Send to WeCom

![Department Efficiency Statistics & Send to WeCom](images/04-department-efficiency-wecom.gif)

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

#### More Examples

![ShotGrid MCP Server Demo](images/sg-mcp.gif)

## Features

| Category | Highlights |
|----------|------------|
| **40+ Tools** | Complete CRUD operations, batch processing, thumbnails, notes, playlists |
| **Transport** | stdio (local), HTTP (remote), ASGI (production) |
| **Performance** | Connection pooling, schema caching, lazy initialization |
| **Skills** | Bundled Agent Skills over MCP (`io.modelcontextprotocol/skills`) |
| **Deployment** | FastMCP Cloud, Docker, uvicorn/gunicorn, any ASGI server |
| **Platform** | Windows, macOS, Linux |

## Quick Start

### Installation

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

# Or using pip
pip install shotgrid-mcp-server
```

### Configuration

Set your ShotGrid credentials:

```bash
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.

```bash
uvx shotgrid-mcp-server
```

#### HTTP Transport - For Remote Access

```bash
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

```json
{
  "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

```json
{
  "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)

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

## Deployment

| Method | Command / Setup |
|--------|-----------------|
| **FastMCP Cloud** | Deploy via [fastmcp.cloud](https://fastmcp.cloud) with `fastmcp_entry.py` |
| **ASGI** | `uvicorn shotgrid_mcp_server.asgi:app --host 0.0.0.0 --port 8000` |
| **Docker** | See [Deployment Guide](docs/deployment.md) |

See the [Deployment Guide](docs/deployment.md) for detailed instructions.

## Available Tools

This server provides **40+ tools** for interacting with ShotGrid:

| Category | Tools |
|----------|-------|
| **CRUD** | `create_entity`, `find_one_entity`, `search_entities`, `update_entity`, `delete_entity` |
| **Batch** | `batch_create`, `batch_update`, `batch_delete` |
| **Media** | `download_thumbnail`, `upload_thumbnail` |
| **Notes** | `shotgrid.note.create`, `shotgrid.note.read`, `shotgrid.note.update` |
| **Playlists** | `create_playlist`, `find_playlists` |
| **MCP Apps** | `shotgrid_dashboard` (interactive status dashboard) |
| **Direct API** | `sg.find`, `sg.create`, `sg.update`, `sg.batch`, and more... |

## MCP Apps

`shotgrid_dashboard` is an [MCP App](https://modelcontextprotocol.io/extensions/apps/build): 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](docs/guide/mcp-apps.md) for the wire format and rebuild instructions.

```bash
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

```bash
# 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](docs/) directory for detailed documentation.

## Contributing

Contributions welcome! Please follow the [Google Python Style Guide](https://google.github.io/styleguide/pyguide.html), write tests, and use [Conventional Commits](https://www.conventionalcommits.org/) — commit messages drive versioning and the changelog. See [RELEASE.md](docs/guide/RELEASE.md) for how releases are cut.

## License

[MIT](LICENSE)

## Architecture

```mermaid
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
```