Installation#
You can run PyAEDT-MCP in two modes:
Start the MCP server and let it launch AEDT later through tools.
Start or reuse an AEDT gRPC session, and then connect PyAEDT-MCP to it.
Check prerequisites#
Python 3.11 or later
AEDT 2022 R2 or later for gRPC workflows
A local AEDT installation or a reachable remote AEDT endpoint
Run PyAEDT-MCP with uvx#
Use uvx when you want to run PyAEDT-MCP without cloning the repository:
uvx --from git+https://github.com/ansys/pyaedt-mcp.git ansys-aedt-mcp
Install PyAEDT-MCP with pip#
Use pip when you want to install PyAEDT-MCP in your current Python
environment:
pip install git+https://github.com/ansys/pyaedt-mcp.git
Install PyAEDT-MCP from source#
Use this option when you want to install from a local clone of the PyAEDT-MCP repository.
git clone https://github.com/ansys/pyaedt-mcp
cd pyaedt-mcp
pip install .
If you plan to contribute, follow the development setup in Contribute for an editable installation, development dependencies, tests, and linters.
Start AEDT in gRPC mode#
If you want to connect to an existing AEDT session, start AEDT in gRPC mode. You can start it from the command line or from the AEDT GUI. PyAEDT supports gRPC in AEDT 2022 R2 and later.
"C:\Program Files\ANSYS Inc\v261\AnsysEM\ansysedt.exe" -grpcsrv 50051
Start PyAEDT-MCP#
PyAEDT-MCP uses STDIO by default.
ansys-aedt-mcp
You can use these common options:
ansys-aedt-mcp --connect --machine localhost --port 50051
ansys-aedt-mcp --transport http --http-host 127.0.0.1 --http-port 8080
ansys-aedt-mcp --include-context
ansys-aedt-mcp --dynamic-tool-discovery
After PyAEDT-MCP starts, connect to it with any MCP-compatible client, such as Claude Code, Copilot CLI, Codex, or Cursor. For more information, see IDE and client configuration.
Review important CLI options#
Option |
Purpose |
|---|---|
|
Choose how PyAEDT-MCP connects to the client. |
|
Point the startup connection to an AEDT gRPC endpoint. |
|
Connect to AEDT during server startup instead of waiting for a tool call. |
|
Register optional guidance tools for AEDT and PyAEDT workflows. |
|
Hide AEDT-only tools until |
|
Set the HTTP transport endpoint. |
Follow a typical first workflow#
After your client connects to PyAEDT-MCP, follow this typical workflow:
Call
check_aedt_installed.Call
check_aedt_status.Use
launch_aedtorconnect_to_aedt.Open a project or create a design.
Use dedicated tools or
run_python_codefor the rest of the workflow.
Note
If the tool surface does not support an operation, the AI agent can generate
a Python snippet and run it with run_python_code in the persistent
PyAEDT session.
For recommended usage patterns, see Best practices. To review the available tools, see Tools and capabilities.