Skip to main content
Glama

zont-mcp

README на русском

A local MCP server that gives AI assistants natural-language control over ZONT heating systems (BAXI Connect+ and compatible controllers) — reading state, switching modes, boiler telemetry, diagnostics and configuration writes.

Demo

A conversation once the server is connected to any MCP client (Claude, Cursor, VS Code, …):

You:      Any boiler errors right now? If yes — what should I do?
Assistant: (calls get_boiler_errors)
          No active faults. Fault flags: none.
          Telemetry: flow temp 54.2 °C, setpoint 55 °C, modulation 34%, status: ch, flame.

You:      We're cold. Run a diagnosis.
Assistant: (calls diagnose_heating)
          Status: warning. The "Boiler circuit" has no active mode — it was
          lost after a controller reboot. Fix: enable the mode.
          (calls set_heating_mode, then get_device_state to verify)
          Done — mode "Comfort" is active, target 21.5 °C.

How it works:

┌─────────────┐   MCP (stdio)   ┌──────────┐   HTTPS    ┌─────────────────┐
│ AI assistant │ ◄──────────────► │ zont-mcp │ ─────────► │  ZONT cloud API │
└─────────────┘                  └──────────┘            └────────┬────────┘
                                                                     │
                                                            ┌────────▼────────┐
                                                            │  ZONT controller │
                                                            │  + boiler (bus)  │
                                                            └─────────────────┘

The server runs locally, opens no network ports and talks to exactly two official ZONT endpoints over HTTPS (my.zont.online/api and lk.zont-online.ru/api).

Related MCP server: unofficialMCP4Vaillant

Quick start

Requirements: Python 3.11+, a ZONT account (zont-online.ru) with a registered device, any MCP-compatible client.

pip install zont-mcp        # from PyPI (Python Package Index)
# — or from source —
git clone https://github.com/MAB2908/zont-mcp.git && cd zont-mcp
python -m venv .venv && source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -e .

Get your API token: set ZONT_LOGIN + ZONT_PASSWORD, run zont-mcp once — the token is printed to stderr; then keep only the token.

Add to your MCP client (Claude Desktop claude_desktop_config.json shown; Cursor, VS Code and others use the same fields — see docs/setup.md for all snippets):

{
  "mcpServers": {
    "zont": {
      "command": "zont-mcp",
      "env": {
        "ZONT_CLIENT_EMAIL": "your@email.com",
        "ZONT_TOKEN": "your_token"
      }
    }
  }
}

Ask your assistant:

Show the list of my ZONT devices.

Credentials live only in environment variables / a local .env file — never in code. Full installation and configuration guide: docs/setup.md (по-русски).

Features

  • 38 tools in 7 groups: device state, heating control, I/O and guard zones, history, analytics, configuration, commands & archives

  • Boiler OpenTherm telemetry — flow temperature, setpoint, burner modulation, status flags

  • Diagnostics with recommendations — known failure signatures: lost bus connection, mode reset after reboot, low pressure, dead radio sensors

  • Configuration writes at web-cabinet level — circuits, modes, sensors, weather-compensation curve; every write is backed up and verified by re-reading

  • Works with any MCP client — Claude Desktop, Claude Code, Cursor, VS Code, ZCode

  • Safety by design — dangerous firmware commands (restart, deletion, SMS/calls) deliberately excluded; device passwords masked in responses

Comparison

The closest analogue is defanator/zont_api — a Python library for the ZONT API. zont-mcp used it as a knowledge source for the z3k_boiler_adapter data format.

Aspect

defanator/zont_api

zont-mcp

Purpose

Python library for developers

MCP server for AI assistants

Interface

import zont_api in code

Natural language via any MCP client

Boiler control

No

Yes — modes, temperatures, relays

Configuration writes

No

Yes — circuits, modes, sensors, weather curve

Radio sensor history

No

Yes

Diagnostics & analytics

No

Yes

Event journal

No

Yes

Prometheus exporter / CSV

Yes

No

Tests

pytest, tox, CI

unittest + GitHub Actions

Choose zont-mcp when you want to talk to your heating system; choose zont_api when you build your own Python software against the API.

Documentation & more examples

Document

Purpose

docs/setup.md

Full installation guide: token, all MCP clients, troubleshooting (по-русски)

AGENTS.md

Working rules for AI agents (English + Russian)

CAPABILITIES.md

Everything the server can and cannot do; z3k command dictionary

examples/heating-check/

Example: full diagnostic session

examples/vacation-mode/

Example: "leaving for a week" scenario

examples/python-client/

Example: use the API client directly from Python

CHANGELOG.md

Release history (English + Russian)

SECURITY.md

Credential handling, vulnerability reporting

Badges

PyPI Python 3.11+ License: MIT Tests MCP

Contributing

See CONTRIBUTING.md — how to set up the project, run tests and submit changes. Simple starter tasks are marked good first issue.

License

MIT — see LICENSE. This is an independent project, not affiliated with ZONT or Micro Line; it uses only the official public ZONT API.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      An MCP server that enables AI assistants to control Home Assistant via natural language, including device control, automation management, and system configuration.
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      An unofficial MCP (Model Context Protocol) server that exposes a Vaillant heat pump's data to AI assistants like Claude. Query outdoor and room temperatures, hot water status, energy consumption, COP estimates, schedules, and diagnostics through natural conversation.
      4
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      MCP server for integrating AI assistants with VkusVill services, allowing natural language access to VkusVill data and operations.
      1
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      An MCP server that lets an AI assistant manage your Keenetic router in plain language — list connected devices, pin static DHCP leases, rename devices, check WAN status, and reboot.
      10
      MIT