shotgrid
A Model Context Protocol (MCP) server for Autodesk ShotGrid/Flow Production Tracking (FPT) with comprehensive CRUD operations and data management capabilities.
claude mcp add loonghao-shotgrid-mcp-server
How to use
ShotGrid MCP Server provides a high-performance bridge that lets AI assistants access Autodesk ShotGrid (Flow Production Tracking) data via the MCP protocol. It exposes 40+ tools for CRUD operations, batch processing, media handling, notes, playlists, and direct API access to ShotGrid, allowing assistants like Claude, Cursor, and VS Code Copilot to query projects, tasks, assets, and timelogs, visualize data, and automate workflows. You can interact through stdio (local) or HTTP (remote) transports, with deployment options including FastMCP Cloud or standard ASGI servers. To use it, install the server, configure your ShotGrid credentials, and connect from an MCP client using the uvx transport (for local usage) or the http transport for remote access.
Once connected, you can issue prompts such as querying task schedules, creating asset batches, or summarizing timelogs, and the MCP server will translate those prompts into ShotGrid API calls via its 40+ tools. The available tools cover CRUD operations (create_entity, find_one_entity, search_entities, update_entity, delete_entity), batch operations (batch_create, batch_update, batch_delete), media handling (download_thumbnail, upload_thumbnail), notes (shotgrid.note.create, shotgrid.note.read, shotgrid.note.update), playlists (create_playlist, find_playlists), and direct API access (sg.find, sg.create, sg.update, sg.batch, and more).
How to install
Prerequisites:
- Python 3.8+ and pip
- Access to ShotGrid with valid credentials
Installation steps:
- Install the MCP server package using pip (recommended):
uv install shotgrid-mcp-server
Alternatively, if you prefer plain pip:
pip install shotgrid-mcp-server
- Verify installation:
python -m pip show shotgrid-mcp-server
- Prepare ShotGrid credentials by exporting environment variables (example):
export SHOTGRID_URL="https://your-site.shotgunstudio.com"
export SHOTGRID_SCRIPT_NAME="your_script_name"
export SHOTGRID_SCRIPT_KEY="your_script_key"
- Run the server using the recommended stdio transport (for local use with Claude Desktop, Cursor, etc.):
uvx shotgrid-mcp-server
- Optional: expose HTTP transport for remote access:
uvx shotgrid-mcp-server http --host 0.0.0.0 --port 8000
- If you are embedding in an MCP client configuration, reference the server as shown in the MCP Client Configuration section of the README.
Additional notes
Tips and notes:
- The default transport is stdio for local use; switch to HTTP transport when remote access is required.
- Ensure SHOTGRID_URL, SHOTGRID_SCRIPT_NAME, and SHOTGRID_SCRIPT_KEY are kept secret and securely stored in your environment or CLI configuration.
- The server supports multiple deployment methods (FastMCP Cloud, ASGI servers like Uvicorn/Gunicorn). See Deployment Guide in the documentation for detailed steps.
- For Claude Desktop or Cursor, configure the MCP client with command uvx and argument shotgrid-mcp-server; for remote clients, provide a proper URL and transport type in the client config.
- If you encounter issues connecting to ShotGrid, verify network access, API permissions, and that the ShotGrid script key has the necessary scopes.
- You can customize or extend tools; they map to ShotGrid REST endpoints and internal abstractions like 40+ tools across CRUD, batch, media, notes, and playlists.
Related MCP Servers
mcp-vegalite
MCP server from isaacwasserman/mcp-vegalite-server
github-chat
A Model Context Protocol (MCP) for analyzing and querying GitHub repositories using the GitHub Chat API.
nautex
MCP server for guiding Coding Agents via end-to-end requirements to implementation plan pipeline
pagerduty
PagerDuty's official local MCP (Model Context Protocol) server which provides tools to interact with your PagerDuty account directly from your MCP-enabled client.
futu-stock
mcp server for futuniuniu stock
mcp -boilerplate
Boilerplate using one of the 'better' ways to build MCP Servers. Written using FastMCP