Key Points
- Initial AI‑generated dashboard was incomplete and hard to use.
- Claude helped migrate about seventy percent of devices to Home Assistant.
- Direct API access via the ha‑mcp add‑on allowed Claude to edit configurations efficiently.
- Automations were created for energy savings and solar monitoring.
- A clean, customized dashboard was built in minutes with natural‑language prompts.
- User supervision was required to approve each change, preventing unwanted actions.
- The entire process reduced weeks of manual work to roughly four hours.
Background and Frustration
The author, a long‑time smart‑home reviewer, manages a sprawling ecosystem that includes Amazon Alexa, Google Home, Apple Home, Samsung SmartThings, Homey, and dozens of bridges such as Lutron Caseta, Philips Hue, Aqara, Ikea and Aeotec. Multiple protocols and platforms left the home feeling like a “Mary Shelley” creation rather than a seamless, Jetsons‑style environment. Existing solutions, including the emerging Matter standard, did not resolve the interoperability gaps for legacy devices.
Exploring Claude for a Solution
Inspired by community projects that used Claude Code to build AI‑driven smart‑home tools, the author asked Claude to generate a universal dashboard. The initial output was a locally hosted web interface that displayed many devices but lacked control options, clear naming and support for bridged hardware. Subsequent attempts to improve the dashboard still produced a cluttered and partially functional result.
Shifting to Home Assistant
Recognizing Home Assistant’s broad compatibility and local execution, the author consulted Claude on integrating the platform. Using natural‑language prompts, Claude guided the discovery of network devices, assisted with adding missing integrations, and suggested practical automations. Within an afternoon, roughly seventy percent of the devices were transferred to Home Assistant, and key automations—such as closing shades when the air conditioner turns on and sending alerts when solar production drops—were created.
Direct API Access via ha‑mcp
To streamline the workflow, the author installed the unofficial ha‑mcp server, a community add‑on that provides Claude direct API access to Home Assistant’s configuration files. This eliminated the slow, browser‑based navigation Claude previously used. With read‑only default permissions and required user approvals for any changes, the setup maintained safety while granting Claude the ability to edit YAML files, install add‑ons and write automations.
Dashboard Creation and Customization
Leveraging the direct API connection, Claude rapidly assembled a clean, user‑friendly dashboard focused on essential controls: lights, locks, climate, cameras and solar metrics. The author further refined the layout by prompting Claude to rearrange elements, apply the “Mushroom” design add‑on and create a single toggle for all office lighting—including a Minka‑Aire ceiling fan, Nanoleaf Blocks, Lifx Luna lamp, Hue bulbs and an Elgato Key Light. Minor troubleshooting, such as resolving a fan‑light conflict, was handled through additional prompts.
Supervision and Limitations
Throughout the process Claude occasionally made errors, such as deleting dashboard sections or misidentifying devices. Each action required manual approval, ensuring the user retained control. While Claude could not yet manage certain HomeKit devices, the overall experience demonstrated a dramatic reduction in setup time—from weeks of trial‑and‑error to roughly four hours of supervised AI interaction.
Implications for the Future
The author notes that while major smart‑home brands have introduced AI chatbots, they lack the depth to troubleshoot complex configurations. Claude’s ability to read logs, suggest automations and generate dashboards points to a future where AI removes the barrier between desire and implementation in smart homes. Home Assistant’s team is exploring deeper AI integration, which could further simplify advanced setups.
Source: theverge.com