AI Models & Platforms

Google Opens Home MCP Early Access to AI Agents for Smart Home Control

mm
Add Unite.AI to your preferred sources on Google

Google opened early access to Home MCP on September 16, 2026, a Model Context Protocol server that lets AI agents monitor devices, review event history, and execute control actions across the Google Home ecosystem. Access is rolling out in English to Google Home Premium Advanced users in the US.

The announcement on the Google Home & Nest Community blog, signed by Taylor, Group Product Manager for Google Home & Nest, states that any agent that supports calling MCP tools (Google named Google Antigravity, Claude, Hermes, and Open Claw as examples) can interact with all devices and event history in the ecosystem, from the Google Nest doorbell and thermostat to Works with Google Home or Matter light bulbs.

According to Google, the connection gives an agent real-world physical context, letting it analyze and answer questions about home activity, summarize camera footage across rooms, manage devices on the user’s behalf, and build smart home dashboards using everyday language. The company described the early access launch as the next step in unlocking agentic smart homes and invited feedback through its Smart Home for Developers Community.

What Connected Agents Can Do

Google said early testers used the connection in several ways. In one, a user asks what the kids did when they got home from school, and the agent returns a summary drawn from every camera along with the relevant clips. Other examples the company cited include using device state history to track how many loads of laundry were done in a week or how long lights were left on, sending an audio message over a Google Home speaker when an agent finishes an asynchronous or proactive task, and assembling a custom dashboard that presents a view of the smart home tailored to the individual user.

Google’s MCP servers overview page groups the server’s functions into four capability areas: entity enumeration, which lists the rooms and devices in a structure; real-time state monitoring, which queries live device state such as connectivity; device control, which executes parameterized actions; and historical analysis, which queries past states and chronological event logs. The page pairs each area with a sample prompt, such as asking how many lights are in the house, checking whether the home is secured, or asking what happened while the user was out. Google characterizes Home MCP as safe and secure by design, stating it is built to meet the company’s security and privacy standards so agents cannot take sensitive actions on a user’s behalf.

Subscription Requirement and Setup

Early access requires an active Google Home Premium Advanced subscription. The Google Store’s Google Home Premium page lists the Advanced plan at $20 per month or $200 per year and the Standard plan at $10 per month or $100 per year. The Advanced tier includes 60 days of event-based video history, 10 days of 24/7 video history for cameras and wired doorbells, descriptive notifications, video-history search, event descriptions, and daily summaries. The store page also notes that Nest Aware is now Google Home Premium, with existing Nest Aware subscribers on the Standard plan and Nest Aware Plus subscribers on the Advanced plan.

According to the Home MCP user guide, setup requires an active smart home in Google Home with connected devices, the Advanced subscription, access to a Google Cloud project, and an MCP-compatible AI client such as Google Antigravity, Claude Cowork, or OpenClaw. Configuration involves creating the Cloud project, enabling the Home API, setting up an OAuth consent screen for an external audience, creating a web-application OAuth client ID with redirect URIs for the target AI client, and publishing the app. The server URL is https://home.googleapis.com/mcp with the OAuth scope home.platform.v2, and Google provides step-by-step instructions for Antigravity, Claude Cowork, and OpenClaw. Familiar-face data carries a separate consent requirement: a manager of the structure must explicitly grant access, and the structure must include at least one compatible Nest camera or doorbell with familiar face detection enabled.

A Companion Server for Developers

Google has also introduced Home Developer MCP, a second server aimed at developers. Google’s overview states that Home Developer MCP grounds AI coding tools in verified technical documentation, including complete Home API references and integration guides, the Matter specification, and OpenThread and Thread documentation. The developer server works with the Google Antigravity suite (its CLI, Antigravity 2.0, and IDE) as well as Claude Code, Cursor, and GitHub Copilot in VS Code, and it authenticates through unauthenticated trial paths or project API keys rather than the user OAuth flow that Home MCP requires. Google’s What’s New page describes the pairing as Home MCP letting personal agents interact with the smart home safely while Home Developer MCP delivers technical information directly to coding tools.

Safety Protections and Stated Limitations

Both the announcement and the user guide state that Home MCP enforces rate limits and safety protections, including a prohibition on sensitive actions such as unlocking doors. Google cautions that connecting a real home to an agent gives that agent control over devices on the user’s behalf and that, depending on the agent, the connection can produce unexpected or even undesired behavior. The company directs users to review its developer policies and terms of service.

The guide advises users whose homes are shared with other household members to inform them that the agent can control devices and access home data, and it suggests creating an additional home for development and testing. Agent access can be revoked at any time from the Google Home app or the user’s My Accounts page.

The guide also lists known issues: some traits are marked experimental and may not work as expected, and users may encounter longer-than-expected latency, which Google says it is actively working to optimize. Creating and managing automations through Home MCP is not supported at this time; Google says it plans to add automations support in a future release.

Aiden Cross is an AI-generated strategist at Unite.AI, covering AI product strategy, execution, and the practical challenges of turning experimental models into scalable, market-ready products. His work focuses on how startups and enterprise teams move from prototypes and demos to reliable systems used by real customers.

With a pragmatic and detail-oriented perspective, Aiden analyzes product roadmaps, go-to-market strategies, platform decisions, and organizational trade-offs that determine whether AI initiatives succeed or stall. He pays particular attention to deployment realities, user adoption, infrastructure constraints, and the alignment between technical capability and business value.

Articles authored by Aiden Cross are AI-generated and reviewed by Unite.AI’s editorial team to ensure clarity, accuracy, and responsible coverage of how AI products are built, shipped, and scaled in the real world.