HomeBrain

HomeBrain is an AI-powered household operating system. It connects structured home data, maintenance workflows, and tool-based AI interactions through an MCP interface, with authentication and session management in front of a browser dashboard.

Household information is scattered: maintenance, routines, and the state of the home rarely live in one place. An assistant is useful when it can read that structure and take bounded actions through tools, with a real session behind the request.

I built the household platform: the browser dashboard, authentication and sessions, the household and maintenance workflows, the API tooling, and the MCP interface over Streamable HTTP.

• Browser-based dashboard for household operations
• Authentication and session management
• Structured household data
• Maintenance workflows
• API tooling for household operations
• MCP interface using Streamable HTTP
• Tool-based AI interactions over that interface

1. 01

   Browser dashboard

   The household operating surface people use directly.

   ↓

2. 02

   Auth and session

   Signed-in access so household data and tools are tied to a session.

   ↓

3. 03

   Household data

   Structured records the workflows and tools can read.

   ↓

4. 04

   Maintenance workflows

   Repeatable upkeep paths built on that household data.

   ↓

5. 05

   MCP over Streamable HTTP

   A defined tool surface for AI interactions, exposed through Streamable HTTP.

1. 01An assistant needs structure

   A chat box over an empty home does not help anyone maintain anything. The assistant needs records and workflows to act on.

2. 02Tools need a session

   Household data is personal. An open tool endpoint is the wrong boundary.

3. 03A defined interface for agents

   If every interaction is free-form chat, the system has no reliable operations.

1. 01An assistant needs structure

   I modeled household data and maintenance workflows first, then exposed those operations as tools.

2. 02Tools need a session

   Authentication and session management sit in front of the dashboard and the API so tool use stays attached to a signed-in session.

3. 03A defined interface for agents

   I exposed household operations through MCP over Streamable HTTP so the tool surface is explicit.

• A household product is a data model plus a few workflows people will actually repeat.
• MCP is useful when the tools map to real operations.
• Session boundaries matter as soon as an agent can touch home data.

• MCP
• Streamable HTTP
• Authentication
• Session management
• API tooling

Private build. The platform connects a browser dashboard, signed-in access, maintenance workflows, and an MCP interface. It is not presented as a public production launch.