Name: Litmus MCP Server
Author: litmusautomation

LABSCO SUMMARY

Built for the questions an OT engineer actually asks.

Several tools exist because of how this data behaves rather than how the API is shaped. get_all_tags_status defaults to showing only non-OK tags, because a plant has thousands and you want the broken ones. get_device_connection_status distinguishes stale from no_data, which is the difference between a dead sensor and a dead gateway. And get_device_data_for_inference bundles metadata, per-tag statistics and recent samples in a single payload so a model can reason about a machine without ten round trips.

What it is

Litmus Automation's official server for Litmus Edge. It covers device configuration, tag management, live and historical data, digital twins, Docker deployment on the edge, and the network and system layer around them.

What you get

62 tools across 13 categories, taking structured arguments and returning JSON
DeviceHub: get_devicehub_devices, create_devicehub_device, get_litmusedge_driver_list for the supported drivers such as ModbusTCP, OPCUA and BACnet, and get_device_connection_status for connected / stale / no_data
Tags: get_devicehub_device_tags paginates with limit and offset, create_devicehub_tag, update_devicehub_tag, delete_devicehub_tag, get_current_value_of_devicehub_tag, and get_all_tags_status which defaults to non-OK so problems surface first
Time series: get_historical_data_from_influxdb, list_influxdb_measurements, query_tag_data, get_tag_statistics for mean, min, max, stddev and count with a baseline range, and get_device_data_for_inference which packs metadata, tags, stats and recent samples into one payload
NATS topics: list_nats_topics, get_current_value_from_topic, get_multiple_values_from_topic for trend analysis
Network and system: get_firewall_rules, get_network_interface_info, start_packet_capture, stop_packet_capture, get_packet_capture_status, get_system_events, get_system_event_stats
Digital twins and Docker: list_digital_twin_models, create_digital_twin_model, get_all_containers_on_litmusedge, run_docker_container_on_litmusedge

Requirements

Docker: docker run -d --name litmus-mcp-server -p 8000:8000 ghcr.io/litmusautomation/litmus-mcp-server:latest. The image is built for linux/amd64, so on ARM64 pass --platform linux/amd64. Port 8000 serves both transports — /mcp for Streamable HTTP and /sse for older clients. HTTPS is available through a TLS-terminating reverse proxy or native TLS with your own certificate. A stdio path exists for Claude Desktop, installed from a clone with uv.

Setup effort

One command plus credentials — docker run -d --name litmus-mcp-server -p 8000:8000 ghcr.io/litmusautomation/litmus-mcp-server:latest, then supply credentials

See installation options

USE CASES

01

Finding which devices stopped reporting

Check connection status across devices and separate stale data from no data at all.

get_devicehub_devicesget_device_connection_status

02

Reading what a sensor says right now

Look up the tag and pull its current value, or subscribe to the topic for the next message.

get_current_value_of_devicehub_tagget_current_value_from_topic

03

Deciding whether a reading is abnormal

Get mean, min, max, stddev and a baseline range for the tag rather than eyeballing a chart.

get_tag_statisticsquery_tag_data

04

Diagnosing a network problem on the edge box

Read firewall rules and interface details, then run a bounded packet capture and collect the file.

get_network_interface_infostart_packet_captureget_packet_capture_status

62 tools

Listed in project docs. Your client may show a different list.

get_litmusedge_driver_list

get_devicehub_devices

create_devicehub_device

get_device_connection_status

get_devicehub_device_tags

get_current_value_of_devicehub_tag

create_devicehub_tag

update_devicehub_tag

delete_devicehub_tag

get_tag_status

get_all_tags_status

get_litmusedge_friendly_name

set_litmusedge_friendly_name

get_cloud_activation_status

get_all_containers_on_litmusedge

BEFORE YOU INSTALL

What to know first

Tools you get

62 tools named in the project's documentation. All 62 register, but some need extra credentials: NATS subscribe tools need the Access Control token, InfluxDB tools need port 8086 and DataHub credentials, and the 16 lem_* tools need an Edge Manager tenant and EDGE_API_TOKEN. Stdio needs ENABLE_STDIO=true.

Requires

A Litmus Edge instance — its base URL with https:// (EDGE_URL) and OAuth2 client credentials from Litmus Edge (EDGE_API_CLIENT_ID, EDGE_API_CLIENT_SECRET)
A linux/AMD64 host — on ARM64, run the Docker image with --platform linux/amd64
Python 3.12+ for the stdio route from a clone
For the topic-value tools, an access token from Litmus Edge → System → Access Control → Tokens (NATS_PASSWORD)
For the historical-data tools, TCP port 8086 allowed in the Litmus Edge firewall and DataHub user credentials (INFLUX_USERNAME, INFLUX_PASSWORD)
For the Litmus Edge Manager tools, an API token issued by LEM (EDGE_API_TOKEN)
On hosts without GitHub access, the litmus-cli binary installed beforehand (or LITMUS_CLI_PATH pointing at one)

INSTALLATION

Connect your client

Server setup

From source

TERMINAL

git clone https://github.com/litmusautomation/litmus-mcp-server.git
cd litmus-mcp-server

# Using uv
uv sync

# Otherwise
pip install -e .

docker

Docker

TERMINAL

docker run -d --name litmus-mcp-server -p 8000:8000 ghcr.io/litmusautomation/litmus-mcp-server:latest

MCP Clients

Claude Code

CLAUDE CODE · JSON

{
  "mcpServers": {
    "litmus-mcp-server": {
      "type": "http",
      "url": "http://localhost:8000/mcp",
      "headers": {
        "EDGE_URL": "${EDGE_URL}",
        "EDGE_API_CLIENT_ID": "${EDGE_API_CLIENT_ID}",
        "EDGE_API_CLIENT_SECRET": "${EDGE_API_CLIENT_SECRET}",
        "NATS_SOURCE": "${NATS_SOURCE}",
        "NATS_PORT": "${NATS_PORT:-4222}",
        "NATS_PASSWORD": "${NATS_PASSWORD}",
        "INFLUX_HOST": "${INFLUX_HOST}",
        "INFLUX_PORT": "${INFLUX_PORT:-8086}",
        "INFLUX_DB_NAME": "${INFLUX_DB_NAME:-tsdata}",
        "INFLUX_USERNAME": "${INFLUX_USERNAME}",
        "INFLUX_PASSWORD": "${INFLUX_PASSWORD}"
      }
    }
  }
}

Cursor

CURSOR · JSON

{
  "mcpServers": {
    "litmus-mcp-server": {
      "url": "http://<YOUR_MCP_SERVER_IP>:8000/sse",
      "headers": {
        "EDGE_URL": "https://<YOUR_LITMUSEDGE_IP>",
        "EDGE_API_CLIENT_ID": "<YOUR_EDGE_API_CLIENT_ID>",
        "EDGE_API_CLIENT_SECRET": "<YOUR_EDGE_API_CLIENT_SECRET>",
        "NATS_SOURCE": "<YOUR_LITMUSEDGE_IP>",
        "NATS_PORT": "4222",
        "NATS_PASSWORD": "<YOUR_NATS_PASSWORD>",
        "INFLUX_HOST": "<YOUR_LITMUSEDGE_IP>",
        "INFLUX_PORT": "8086",
        "INFLUX_DB_NAME": "tsdata",
        "INFLUX_USERNAME": "<YOUR_INFLUX_USERNAME>",
        "INFLUX_PASSWORD": "<YOUR_INFLUX_PASSWORD>"
      }
    }
  }
}

VS Code

VS CODE · JSON

{
  "servers": {
    "litmus-mcp-server": {
      "url": "http://<YOUR_MCP_SERVER_IP>:8000/sse",
      "headers": {
        "EDGE_URL": "https://<YOUR_LITMUSEDGE_IP>",
        "EDGE_API_CLIENT_ID": "<YOUR_EDGE_API_CLIENT_ID>",
        "EDGE_API_CLIENT_SECRET": "<YOUR_EDGE_API_CLIENT_SECRET>",
        "NATS_SOURCE": "<YOUR_LITMUSEDGE_IP>",
        "NATS_PORT": "4222",
        "NATS_PASSWORD": "<YOUR_NATS_PASSWORD>",
        "INFLUX_HOST": "<YOUR_LITMUSEDGE_IP>",
        "INFLUX_PORT": "8086",
        "INFLUX_DB_NAME": "tsdata",
        "INFLUX_USERNAME": "<YOUR_INFLUX_USERNAME>",
        "INFLUX_PASSWORD": "<YOUR_INFLUX_PASSWORD>"
      }
    }
  }
}

Windsurf

WINDSURF · JSON

{
  "mcpServers": {
    "litmus-mcp-server": {
      "serverUrl": "http://<YOUR_MCP_SERVER_IP>:8000/sse",
      "headers": {
        "EDGE_URL": "https://<YOUR_LITMUSEDGE_IP>",
        "EDGE_API_CLIENT_ID": "<YOUR_EDGE_API_CLIENT_ID>",
        "EDGE_API_CLIENT_SECRET": "<YOUR_EDGE_API_CLIENT_SECRET>",
        "NATS_SOURCE": "<YOUR_LITMUSEDGE_IP>",
        "NATS_PORT": "4222",
        "NATS_PASSWORD": "<YOUR_NATS_PASSWORD>",
        "INFLUX_HOST": "<YOUR_LITMUSEDGE_IP>",
        "INFLUX_PORT": "8086",
        "INFLUX_DB_NAME": "tsdata",
        "INFLUX_USERNAME": "<YOUR_INFLUX_USERNAME>",
        "INFLUX_PASSWORD": "<YOUR_INFLUX_PASSWORD>"
      }
    }
  }
}

Claude Desktop (stdio)

CLAUDE DESKTOP (STDIO) · JSON

{
  "mcpServers": {
    "litmus-mcp-server": {
      "command": "<YOUR_PATH_TO_VENV>/bin/python3",
      "args": [
        "<YOUR_PATH_TO_LITMUS_MCP_SERVER>/src/server.py"
      ],
      "env": {
        "ENABLE_STDIO": "true",
        "PYTHONPATH": "<YOUR_PATH_TO_LITMUS_MCP_SERVER>/src",
        "EDGE_URL": "https://<YOUR_LITMUSEDGE_IP>",
        "EDGE_API_CLIENT_ID": "<YOUR_EDGE_API_CLIENT_ID>",
        "EDGE_API_CLIENT_SECRET": "<YOUR_EDGE_API_CLIENT_SECRET>",
        "NATS_SOURCE": "<YOUR_LITMUSEDGE_IP>",
        "NATS_PORT": "4222",
        "NATS_PASSWORD": "<YOUR_NATS_PASSWORD>",
        "INFLUX_HOST": "<YOUR_LITMUSEDGE_IP>",
        "INFLUX_PORT": "8086",
        "INFLUX_DB_NAME": "tsdata",
        "INFLUX_USERNAME": "<YOUR_INFLUX_USERNAME>",
        "INFLUX_PASSWORD": "<YOUR_INFLUX_PASSWORD>"
      }
    }
  }
}

Other routes

Docker (arm64)

DOCKER (ARM64) · TERMINAL

docker run -d --name litmus-mcp-server --platform linux/amd64 -p 8000:8000 ghcr.io/litmusautomation/litmus-mcp-server:main

The ARM64-specific part is --platform linux/amd64 (the image is built for linux/amd64). The README prints this line with the :main tag — the image built from the main branch — while its other commands use :latest; both tags are published on ghcr.io.

Litmus MCP Server