planpatch
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@planpatchAlex has to stay late until 20:00, how can the household still get everything done?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
PlanPatch
Replan a household's day when someone's work shift changes.
Alex has to stay late at work. Someone still needs to pick up the kids, collect groceries and make dinner. PlanPatch works out who can cover each task, accounts for travel time, and shows what has to move or wait until tomorrow.

Watch the demo — a 98-second walkthrough of the running app.
This is a small prototype for the Alexa+ track of the Amazon Developer Hackathon. It has a local web app and an MCP server; it hasn't been connected to Alexa+ yet.
Run it
You'll need Node.js 22+ and npm. No API key is needed.
npm ci
npm startOpen localhost:4178.
The demo uses a fictional family and four tasks. Try leaving Alex unavailable from 14:00 to 20:00 with no helper budget. Compare the suggested plans, review the changes, then apply one. The saved schedule survives a refresh; Undo last applied plan restores it.
Change the end time to 23:00 to see which optional task gets deferred. Allow Jo as a helper with a $12 budget to compare another option. That price is sample data, and applying a plan only updates local data.
Related MCP server: KinSwitch
MCP
With the server running, open another terminal:
npm run demo:mcpThe script connects through the MCP SDK, reads the household and requests a recovery plan. Use Review assistant proposal in the web app to open it.
The endpoint is http://127.0.0.1:4178/mcp, using Streamable HTTP with MCP 2025-11-25. For clients that accept this configuration:
{
"mcpServers": {
"planpatch": { "url": "http://127.0.0.1:4178/mcp" }
}
}Tool | What it does |
| Reads tasks, availability and the saved schedule |
| Computes alternatives without changing assignments |
| Retrieves a saved proposal |
Applying a plan is handled in the web app after review. Proposals expire after 30 minutes, and an old proposal can't overwrite a newer schedule.
How it works
The planner searches the sample day's time slots and possible assignees. It checks availability, travel time, workload, fixed appointments and the helper budget. Plans are ranked by deferrals, cost, reassignment and time changes. If an essential task can't be covered, it reports that rather than dropping the task.
src/planner.js— scheduling and scoringsrc/store.js— JSON storage, revisions and undosrc/server.js— MCP tools and web routespublic/— plain HTML, CSS and JavaScript
State is stored in .data/, which is ignored by Git.
Tests
npm testTests cover scheduling constraints, impossible plans, helper permissions, persistence, expired or stale proposals, undo, and MCP calls over HTTP. Browser checks are recorded in docs/validation.md.
Current limits
The planner handles one sample day, not overnight or multi-day schedules. Travel times and workload limits are fixed examples. There are no calendar, messaging, payment or booking connections.
The server is for local use and binds to 127.0.0.1. Cloud clients can't reach that address, and the app doesn't have multi-user authentication, so don't expose it publicly as-is.
Dependencies
Uses the MCP TypeScript SDK, Express and Zod.
Related MCP Connectors
Read tasks, habits, events and scheduling links; create and update Reclaim tasks and habits.
- QuietaOAuthapp.getquieta
Family calendar MCP server: read the family's week and propose changes a parent approves.
Family schedules and household tools with OAuth. External calendars remain read-only.
Read-only A2Me family context tools for AI assistants (members, dates, activity, relationships)
Related MCP Servers
- FlicenseNot gradedqualityBmaintenanceEnables personal productivity management by exposing MCP tools to create, retrieve, list, update, complete, and delete tasks and calendar events, with validation and conflict detection.-
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to manage household schedules by checking availability, proposing minimal changes, and requiring explicit human approval before executing any action.MIT
- AlicenseNot gradedqualityCmaintenanceEnables persistent household routine planning with interruption replanning, preserving completed progress and rejecting stale updates through MCP tools over Streamable HTTP.MIT
- AlicenseAqualityBmaintenanceEnables household management through natural language, covering bills, chores, shopping, reminders, budget, and smart-home devices, with every mutating action dry-run, confirmed, and audited.31MIT