Forgeline
Read-only Modbus TCP monitoring server that exposes safe MCP tools for AI agents to read holding/input registers, coils, and device identity from industrial devices without write access.
README
Forgeline
Read-only Modbus TCP monitoring, exposed as safe MCP tools.
<p align="center"> <img src="docs/architecture.svg" alt="Forgeline architecture: an AI agent reads a Modbus device over MCP, read-only by design" width="100%"> </p>
Forgeline is an MCP server that lets an AI agent monitor an industrial Modbus TCP device — reading holding/input registers, coils, and device identity — without being able to change anything.
It's read-only on purpose: there are no write or command tools, and no code that writes coils or registers, so a connected agent can't alter device state.
A simulated Modbus device is bundled, so you can run it with no hardware.
Status
Working MVP. Four read-only tools — get_device_info, read_holding_registers,
read_input_registers, read_coils — tested against the bundled simulator.
Project layout
forgeline/
├── src/forgeline/
│ ├── server.py # FastMCP server: registers MCP tools
│ └── tools.py # Read-only Modbus implementations (transport-agnostic)
├── simulator/
│ └── device.py # Bundled pymodbus TCP simulator (zero hardware)
├── docker-compose.yml # Starts simulator + server together
├── Dockerfile
├── pyproject.toml
├── README.md
└── LICENSE # Apache 2.0
Quick start
Option A — Docker (simulator + server, one command)
docker compose up --build
This starts:
- simulator — a virtual Modbus device on
localhost:5020. - forgeline — the MCP server over
streamable-httpathttp://localhost:8000/mcp, already pointed at the simulator.
Option B — Local Python
Requires Python 3.10+.
pip install -e .
# Terminal 1: start the bundled simulator
forgeline-simulator
# Terminal 2: run the MCP server (stdio transport, for an MCP client)
forgeline
By default the server talks to the simulator on 127.0.0.1:5020.
Configuration
All configuration is via environment variables.
| Variable | Default | Description |
|---|---|---|
MODBUS_HOST |
127.0.0.1 |
Host of the Modbus TCP device to monitor. |
MODBUS_PORT |
5020 |
Port of the Modbus TCP device. |
MODBUS_UNIT_ID |
1 |
Modbus unit / slave id. |
MODBUS_TIMEOUT |
3.0 |
Per-request timeout in seconds. |
FORGELINE_TRANSPORT |
stdio |
MCP transport: stdio, sse, or streamable-http. |
FORGELINE_HOST |
127.0.0.1 |
Bind host for HTTP transports. |
FORGELINE_PORT |
8000 |
Bind port for HTTP transports. |
SIMULATOR_HOST |
0.0.0.0 |
Bind host for the bundled simulator. |
SIMULATOR_PORT |
5020 |
Bind port for the bundled simulator. |
Connecting an MCP client
To launch Forgeline over stdio (e.g. from Claude Desktop or another MCP client),
register a server that runs the forgeline command. Example client config:
{
"mcpServers": {
"forgeline": {
"command": "forgeline",
"env": { "MODBUS_HOST": "127.0.0.1", "MODBUS_PORT": "5020" }
}
}
}
(Start forgeline-simulator first, or point MODBUS_HOST/MODBUS_PORT at a
real device.)
Tools
get_device_info
Returns identity and connection details for the monitored device.
{
"host": "127.0.0.1",
"port": 5020,
"unit_id": 1,
"connected": true,
"device_identification_supported": true,
"identity": {
"vendor_name": "Forgeline",
"product_code": "FL-SIM-1",
"revision": "1.0.0",
"vendor_url": "https://github.com/codenikhildr/Forgeline",
"product_name": "Forgeline Modbus Simulator",
"model_name": "Virtual PLC",
"user_application_name": "Forgeline MVP Simulator"
}
}
Devices that do not implement Modbus device identification still return
connection details, with device_identification_supported: false.
read_holding_registers / read_input_registers
Read 16-bit registers — holding (FC 0x03, read/write on the device) or input (FC 0x04, read-only measurements). Parameters:
| Param | Type | Default | Bounds |
|---|---|---|---|
address |
int | — | 0–65535 |
count |
int | 1 |
1–125 (Modbus FC03/04 limit) |
// read_holding_registers(address=0, count=5)
{
"start_address": 0,
"count": 5,
"unit_id": 1,
"values": [
{ "address": 0, "value": 100 },
{ "address": 1, "value": 101 },
{ "address": 2, "value": 102 },
{ "address": 3, "value": 103 },
{ "address": 4, "value": 104 }
]
}
read_coils
Read single-bit on/off coils (FC 0x01). Parameters:
| Param | Type | Default | Bounds |
|---|---|---|---|
address |
int | — | 0–65535 |
count |
int | 1 |
1–2000 (Modbus FC01 limit) |
// read_coils(address=0, count=4)
{
"start_address": 0,
"count": 4,
"unit_id": 1,
"values": [
{ "address": 0, "value": false },
{ "address": 1, "value": true },
{ "address": 2, "value": false },
{ "address": 3, "value": true }
]
}
Requests outside these bounds (count over the limit, or a range that exceeds the 16-bit address space) are rejected before any request reaches the device.
Simulator data map
The bundled simulator pre-populates each table (addresses are zero-based):
| Table | FC | Addresses | Values |
|---|---|---|---|
| Holding registers | 0x03 | 0–99 | 100 + addr |
| Input registers | 0x04 | 0–99 | 200 + addr |
| Coils | 0x01 | 0–99 | addr % 2 |
| Discrete inputs | 0x02 | 0–99 | (addr+1) % 2 |
Development
Run the test suite — it starts the bundled simulator on a test port and exercises all four tools plus the input-validation guards:
pip install -e ".[dev]"
pytest
CI runs the same suite on every push and pull request across Python 3.10–3.12.
License
Apache License 2.0. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.