Skip to main content
Glama

glasshome-mcp

MCP sidecar for GlassHome: exposes GlassHome's SQLite database and Home Assistant entity control as Model Context Protocol tools. Runs inside the existing GlassHome container — no separate container, no extra env vars.

What it does

Any MCP-compatible LLM client (Hermes Agent, Claude Desktop, etc.) can:

  • Read HA entities — list, filter, and get states for any entity

  • Control HA entities — turn lights on/off, adjust climate, activate scenes

  • Edit dashboards — add/remove/rearrange widgets, create dashboards

  • Install widgets — download from the GlassHome Hub API, cache bundles, register in DB

  • Query the GlassHome DB — dashboards, widgets, accounts, settings

Related MCP server: hass-mcp-server

Architecture

GlassHome container (ghcr.io/glasshome/dash:beta)
├── dash-server     :3123  (Bun, s6-supervised)
├── glasshome-mcp   :3900  (Bun, MCP Streamable HTTP)
├── /data/glasshome.sqlite
└── /data/bundles/
         │
         ▼
  Home Assistant (HTTPS)

The MCP server is a child process started by /run.sh before the main dash-server execs. Both share the same SQLite DB and bundle directory — no volume sharing or sidecar containers.

HA Auth

No separate HA_TOKEN env var needed. The MCP server reads HA URL + OAuth tokens directly from GlassHome's SQLite account table (provider_id='hass'). Tokens auto-refresh via /auth/token.

Quick start

1. Deploy

./setup.sh deploy your-host /path/to/compose.yaml

Builds the image, transfers it, loads it, and recreates the GlassHome container. MCP is live at http://your-host:3900/mcp.

2. Update your compose file

In your existing GlassHome service, point the image at glasshome-mcp:latest and add port 3900:

services:
  glasshome:
    image: glasshome-mcp:latest   # was: ghcr.io/glasshome/dash:beta
    ports:
      - "3123:3123"               # GlassHome UI (existing)
      - "3900:3900"               # MCP sidecar (new)

No HA_TOKEN needed — the MCP reads HA URL + OAuth tokens from GlassHome's SQLite DB automatically.

3. Connect your MCP client

Point any MCP client at http://<host>:3900/mcp. The server speaks MCP Streamable HTTP.

Tools (20)

Category

Tools

Entities (7)

list_entities, get_entity_states, set_entity_state, get_entity, list_areas, call_service, get_device_info

Dashboards (4)

get_dashboard, get_dashboard_summary, create_dashboard, delete_dashboard

Widgets (7)

add_widget, remove_widget, update_widget, move_widget, list_widgets, describe_widget_schema, get_widget

Install (2)

install_widget, search_widgets

Requirements

  • GlassHome container (ghcr.io/glasshome/dash:beta or later)

  • Docker on the host

  • HA URL reachable from the container (auto-detected from household_settings.hass_url_default)

License

MIT

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
      B
      maintenance
      An MCP server and Home Assistant add-on that enables AI assistants to manage smart homes by creating automations, designing dashboards, and interacting with entities. It features native access to Home Assistant APIs, built-in Git versioning for safe rollbacks, and full management of HACS integrations.
      657
      MIT
    • A
      license
      B
      quality
      D
      maintenance
      MCP server for full Home Assistant control, enabling AI agents to manage dashboards, automations, files, apps, entities, and more via REST API, WebSocket, and SSH.
      66
      182 npm
      MIT
    • A
      license
      A
      quality
      A
      maintenance
      A comprehensive MCP server that enables AI assistants to interact with Home Assistant, control smart home devices, query states, execute services, and manage automations via natural language.
      77
      186,916 PyPI
      7,902
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      A Model Context Protocol (MCP) server for Home Assistant integration with Claude and other LLMs, enabling querying and controlling Home Assistant entities, automations, and dashboards.
      29
      MIT