EPICS-MCP-Server
Overview
The EPICS MCP Server is a Python-based server designed to interact with EPICS (Experimental Physics and Industrial Control System) process variables (PVs). It provides a set of tools to retrieve PV values, set PV values, and fetch detailed information about PVs. The server is built using the mcp framework and communicates over stdio, making it suitable for integration into larger control systems or workflows.
This tool is particularly useful in environments where EPICS PVs are used for monitoring and controlling hardware or software parameters.
Features
- The EPICS MCP Server provides the following tools:
get_pv_value
- Create or update a single file in a repository
- Inputs:
pv_name(string): The name of the PV variable.
- Returns: A JSON object containing the status (
successorerror) and the retrieved value or an error message.
set_pv_value
- Set a new value for a specified PV.
- Inputs:
pv_name(string): The name of the PV variable.pv_value(string): The new value to be set for the PV.
- Returns: A JSON object containing the status (
successorerror) and a confirmation message or an error message.
get_pv_info
- Fetches detailed information about a specified PV.
- Inputs:
pv_name(string): The name of the PV variable.
- Returns: A JSON object containing the status (
successorerror) and the detailed information about the PV or an error message.
Usage with Langchain
- To use this with Langchain, you must install the dependencies required for the project.
pip install -r requirements.txt
Langchain
server_params = StdioServerParameters(
command="python",
# Make sure to update to the full absolute path to your math_server.py file
args=["/path/server.py"],
)
EPICS
- Before using the EPCIS mcp server, you must successfully install EPCIS on your local machine, ensure that IOC can start normally, and verify that functions such as
caget,caput, andcainfoare working properly. For detailed installation instructions, please refer to https://epics-controls.org/resources-and-support/base/.
jiangyan@DESKTOP-84CO9VB:~$ softIoc -d ~/EPICS/DB/test.db
Starting iocInit
############################################################################
## EPICS R7.0.8
## Rev. 2025-02-13T14:29+0800
## Rev. Date build date/time:
############################################################################
iocRun: All initialization complete
epics>
jiangyan@DESKTOP-84CO9VB:~$ caget temperature:water
temperature:water 88
jiangyan@DESKTOP-84CO9VB:~$ caput temperature:water 100
Old : temperature:water 88
New : temperature:water 100
jiangyan@DESKTOP-84CO9VB:~$ cainfo temperature:water
temperature:water
State: connected
Host: 127.0.0.1:5056
Access: read, write
Native data type: DBF_DOUBLE
Request type: DBR_DOUBLE
Element count: 1
Test Result
- Mcp client:
async def run():
async with stdio_client(server_params) as (read, write):
async with ClientSession(read, write) as session:
# Initialize the connection
await session.initialize()
# Get tools
tools = await load_mcp_tools(session)
# Create and run the agent
agent = create_react_agent(model, tools)
agent_response = await agent.ainvoke({"messages": "To query the value of a PV (Process Variable) named temperature:water"})
return agent_response
)
- Result:
================================[1m Human Message [0m=================================
To query the value of a PV (Process Variable) named temperature:water
==================================[1m Ai Message [0m==================================
Tool Calls:
get_pv_value (call_vvbXwi51CyYUxEM0hcyvCFCY)
Call ID: call_vvbXwi51CyYUxEM0hcyvCFCY
Args:
pv_name: temperature:water
=================================[1m Tool Message [0m=================================
Name: get_pv_value
{
"status": "success",
"value": 88.0
}
==================================[1m Ai Message [0m==================================
The current value of the PV named `temperature:water` is 88.0.
EPICS Control Server
Project Details
- Jacky1-Jiang/EPICS-MCP-Server
- MIT License
- Last Updated: 4/25/2025
Recomended MCP Servers
A Gmail MCP server
An MCP (Model Context Protocol) server for performing accessibility audits on webpages using axe-core. Use the results in...
SSAFY 학생들을 위한 프로젝트 포트폴리오 및 면접 준비 도우미 서버
A demonstration server implementing the Model Context Protocol (MCP)
A lightweight, easy-to-use Model Context Protocol (MCP) implementation that seamlessly integrates Perplexity's powerful AI models into Claude Desktop....
MCP Server for Shopify API
This is open base storage and it is mainly used for a lots of proxy tools like Clash,...





