How to Build a Raspberry Pi Home Dashboard with OpenClaw

MagicMirror has mass. The project crossed 19,000 GitHub stars and hosts thousands of community modules covering weather, calendars, transit schedules, Spotify playback, and Home Assistant device status. DAKboard offers a simpler hosted alternative with drag-and-drop layouts and calendar sync. Both solve the "what to display" problem well.

Neither solves the "who manages it" problem. MagicMirror modules pull data on fixed intervals, and the only way to change what's shown is to edit a JavaScript config file over SSH. DAKboard handles layout changes through a web UI, but you still pick which widgets appear and when. If you want the dashboard to show your meeting agenda in the morning, switch to weather radar when a storm is approaching, and surface a grocery list on Saturday afternoon, you're writing custom cron scripts or doing it by hand.

OpenClaw is an open-source AI agent framework that runs on the same Raspberry Pi as your dashboard. It connects to cloud LLMs (Anthropic, OpenAI, Google) for reasoning while keeping the local footprint under 200 MB of RAM. The Pi handles display rendering and agent orchestration. The cloud handles inference. The result is a dashboard that adapts to context without manual config changes.

This guide walks through three stages: picking your dashboard software, installing OpenClaw alongside it, and building automated routines that keep the display relevant throughout the day.

Your dashboard software runs the display. OpenClaw sits behind it, managing what gets shown and when. The two most practical options are MagicMirror and DAKboard, and the choice depends on how much control you want over layout and modules.

MagicMirror is free, open source, and runs locally on the Pi. It renders a full-screen Electron app divided into regions (top bar, upper third, middle center, lower third, bottom bar) where modules slot in. The community has built modules for nearly everything: Google Calendar sync, OpenWeatherMap forecasts, public transit departures, Todoist task lists, Spotify now-playing, Home Assistant entity states, and news feeds via RSS. Configuration lives in a single config.js file. Adding a module means cloning its repo into the modules/ directory and adding a block to the config. The learning curve is real but manageable if you're comfortable editing JavaScript objects.

DAKboard takes the opposite approach. It's a hosted service that renders a dashboard as a web page. You configure it through a browser-based editor, point your Pi's Chromium to the URL in kiosk mode, and the display updates automatically. Calendar integration works with Google, Outlook, and Apple calendars out of the box. Weather, photos, news, and custom HTML widgets are available on the free tier. The Pro plan ($5.99/month) adds scheduling, multiple screens, and custom CSS. DAKboard requires less setup but gives you less control, and it depends on the hosted service staying online.

For OpenClaw integration, MagicMirror is the stronger choice. Its local config file is something OpenClaw can read and modify through shell commands, and modules can be enabled or disabled programmatically. DAKboard's config lives on their servers, so OpenClaw would need to interact through their API or browser automation, which adds fragility.

A third option worth noting: a plain Chromium kiosk pointing at a self-hosted page. If you want to build a fully custom dashboard using HTML, CSS, and JavaScript, OpenClaw can manage it by editing the page files directly. This is the most flexible path but requires the most frontend work.

Both MagicMirror and OpenClaw run on Node.js, and they coexist without conflicts on a Raspberry Pi 5 (4 GB or 8 GB). Here's the setup order.

Step 1: Flash Raspberry Pi OS

Use the official Raspberry Pi Imager to write Raspberry Pi OS (64-bit, Desktop) to your SD card or USB SSD. The Desktop variant includes the window manager needed for MagicMirror's Electron display. Enable SSH during imaging so you can configure remotely.

Step 2: Install MagicMirror

SSH into the Pi and run the MagicMirror installer:

curl -sL https://raw.githubusercontent.com/sdeber/MagicMirror-Module-Installer/master/install.sh | bash

Or clone and install manually:

git clone https://github.com/MagicMirrorOrg/MagicMirror
cd MagicMirror
npm run install-mm
cp config/config.js.sample config/config.js

Edit config/config.js to add your initial modules (clock, weather, calendar). Start it with npm start to verify the display works.

Step 3: Install OpenClaw

In a separate SSH session:

curl -fsSL https://openclaw.ai/install.sh | bash
openclaw onboard --install-daemon

The onboarding wizard walks you through connecting your LLM API keys (Anthropic, OpenAI, or others) and optionally linking a messaging channel like Telegram or Discord. Since this is a dashboard Pi, you might skip the messaging channel and interact through SSH or a local web interface instead.

Step 4: Verify coexistence

With both running, check resource usage:

htop

MagicMirror's Electron process typically uses 200 to 400 MB depending on modules. OpenClaw's gateway uses under 200 MB. On a 4 GB Pi, that leaves comfortable headroom for the OS and background processes. On an 8 GB Pi, resource pressure is a non-issue.

Store and sync your dashboard assets from anywhere

Fastio gives your Raspberry Pi dashboard a cloud workspace for photos, configs, and data exports. generous storage, no credit card, and an MCP endpoint your OpenClaw agent can talk to directly.

Start 14-Day Trial

OpenClaw's built-in cron system is where the dashboard gets intelligent. Instead of showing the same layout all day, you can schedule the agent to swap modules, update data sources, or change display content based on time, day of week, or external triggers.

OpenClaw supports three schedule types: one-shot (--at), fixed interval (--every), and standard cron expressions (--cron). Each job can run in the main session or in an isolated session, and results can be delivered to a messaging channel or webhook.

Morning briefing layout

Create a cron job that runs at 7 AM and tells OpenClaw to configure MagicMirror for a morning-focused display:

openclaw cron create "0 7 * * *" \
  "Switch MagicMirror to morning mode: enable calendar, weather forecast, and news modules. Disable the photo slideshow and Home Assistant module." \
  --name "Morning dashboard" \
  --tz "America/New_York" \
  --session main

OpenClaw interprets the prompt, edits config/config.js to toggle the relevant modules, and restarts MagicMirror. The agent knows how to read and modify the config because it has shell access to the Pi's filesystem.

Evening wind-down layout

A second job at 8 PM shifts the display to something calmer:

openclaw cron create "0 20 * * *" \
  "Switch MagicMirror to evening mode: show photo slideshow, weather for tomorrow, and dim the backlight to 40%." \
  --name "Evening dashboard" \
  --tz "America/New_York" \
  --session main

Weekend override

On weekends, suppress the calendar and show more relaxed content:

openclaw cron create "0 9 * * 6,0" \
  "Weekend mode: disable work calendar, enable family photo slideshow and grocery list widget." \
  --name "Weekend dashboard" \
  --tz "America/New_York" \
  --session main

Event-driven updates

Beyond time-based schedules, OpenClaw's webhook endpoint can trigger dashboard changes from external systems. If Home Assistant detects that a severe weather alert has been issued, it can POST to OpenClaw's hook endpoint, and the agent switches the display to show a full-screen radar widget. The webhook configuration uses a shared secret for authentication and accepts JSON payloads at a local endpoint.

Each cron job consumes LLM tokens when it fires, so keep prompts specific and schedules reasonable. The OpenClaw docs recommend testing manually before scheduling, and running complex jobs in isolated sessions to avoid interfering with your main agent context.

A home dashboard accumulates assets: family photos for slideshows, custom CSS themes, module configuration backups, and data exports from the agent. Local SD card storage works for basics, but it's fragile (SD cards fail) and isolated (you can't update photos from your phone without SSH access).

Fastio provides a cloud workspace that OpenClaw can read from and write to through the MCP server. The Business Trial includes 50 GB of storage, included credits, and 5 workspaces, with no credit card required.

Syncing slideshow photos

Create a workspace on Fastio called "Dashboard Photos." Upload family photos through the web UI, the desktop app, or Cloud Import from Google Drive, Dropbox, OneDrive, or Box. Then schedule OpenClaw to pull new photos to the Pi periodically. The agent can use the Fastio MCP tools to list workspace contents and download files to the local photo directory that MagicMirror's photo module watches.

Configuration backups

MagicMirror's config.js is the single point of failure for your dashboard layout. If the SD card corrupts or you experiment with a change that breaks things, you lose your setup. Schedule OpenClaw to back up the config to a Fastio workspace daily. The agent reads the file, uploads it with a timestamped name, and you have a version history you can restore from.

Shared dashboard for households

If multiple people contribute to the dashboard (one person manages the grocery list widget, another curates photos), Fastio workspaces let everyone upload to the same place. The workspace supports granular permissions, so you can give family members upload access without exposing agent configuration files. OpenClaw polls the workspace for changes and updates the Pi's local files when new content arrives.

For comparison, you could use S3 or Google Drive for the same sync tasks. S3 requires AWS credentials and manual scripting. Google Drive works through rclone but needs periodic re-authentication. Fastio's MCP integration means OpenClaw can interact with the workspace using natural language tool calls rather than custom scripts, and the Intelligence mode auto-indexes uploaded files for search and chat.

The value of an OpenClaw-managed dashboard isn't any single module. It's the ability to compose and switch between module sets based on what's actually useful at a given moment. Here are tested combinations and the problems you'll hit.

Information hub (general purpose)

Clock, weather (MMM-OpenWeatherMap), Google Calendar (default calendar module), news feed (newsfeed module), and a compliments/quotes module for visual variety. This is the default MagicMirror config with minor additions. OpenClaw adds value by scheduling seasonal adjustments: switching the weather module to show a 10-day forecast during travel planning weeks, or adding a countdown module before a family event.

Smart home control panel

Home Assistant entity display (MMM-HomeAssistant), room temperatures, door/window states, and camera feeds. OpenClaw connects to Home Assistant through the community skill on ClawHub, which uses the Assist (Conversation) API. You can ask the agent to change what entities are displayed, or trigger an automation when the dashboard shows an anomaly. The Home Assistant Community forum documents this integration for both add-on and standalone deployments.

Productivity wall display

Todoist tasks, GitHub contribution graph, Pomodoro timer, and a Spotify now-playing widget. Good for a home office. OpenClaw can clear completed tasks from the display, shuffle the playlist based on focus mode, or suppress notifications during deep work blocks.

Troubleshooting common issues

Screen burn-in on LCD panels is a real concern for always-on displays. Use MagicMirror's MMM-ModuleScheduler to rotate module positions periodically, and set the backlight to dim overnight. OpenClaw can handle this: schedule a job that tells the agent to cycle layouts every few hours.

MagicMirror crashes after module updates are common. The Electron process sometimes fails to restart cleanly after a config change. Use a systemd service with Restart=always to auto-recover, and have OpenClaw verify the display is running after each config modification by checking the process list.

SD card wear is the long-term killer for Pi projects. Use a USB SSD instead of an SD card for the boot drive. It costs $15 to $20 for a 128 GB drive and dramatically extends the lifespan of an always-on system. The OpenClaw docs specifically recommend this for persistent deployments.

Wi-Fi drops cause weather and calendar modules to show stale data. Wire the Pi with Ethernet if possible. If Wi-Fi is the only option, OpenClaw can monitor connectivity and display a status indicator when the network is unreachable, so you know the data on screen might be outdated.