LystBot

MCP server examples: 10 personal workflows

These MCP server examples show how an assistant can turn a request into a saved list item or another tool action. For a working list example, authenticate with npx lystbot login YOUR_API_KEY and connect LystBot with claude mcp add lystbot -- npx lystbot mcp.

LystBot tool Personal use Saved result
add_items Meal planning Ingredients in a shared list
create_list Travel packing A new packing list
share_list Household shopping A share code

See the LystBot agent setup for the full reference. Related guides: Claude grocery workflow, saving agent tasks in shared lists.

10 MCP Server use cases for personal productivity - icons showing grocery lists, calendar, terminal, and more

1. MCP server use cases for grocery and shopping lists

MCP server: LystBot MCP

You're chatting with Claude about a recipe. You realize you need ingredients. Instead of switching apps, copying items, opening your phone:

"Add oat milk, tahini, and chickpeas to my grocery list"

Done. Your partner sees the updated list on their phone immediately because LystBot syncs across devices.

LystBot lists overview showing shared grocery and shopping lists LystBot list detail view with grocery items added via MCP

The reason this works so well: grocery lists are the kind of task you do while doing something else. You're meal planning, browsing recipes, or just remembering you're out of coffee. Having your AI assistant handle the list means you stay in flow.

We wrote a full walkthrough of this setup: Using Claude as your MCP-powered grocery assistant.

LystBot is our product. Its documented tools cover lists, sharing, categories, attachments and reminders. If you're curious about the technical side, we wrote about building a list app with an API-first approach.

2. File system management

Potential server: filesystem. Verify its current operations before connecting it.

"Find all PDFs in my Downloads folder that are older than 30 days and move them to an Archive folder."

A file-management workflow needs access to the intended folders. Before trying a move, ask for the matching paths and review them. Keep the first test in a disposable folder.

3. Calendar and scheduling

Potential server: Google Calendar MCP. Verify its current operations before connecting it.

"What is on my calendar next week? Find a two-hour gap on Thursday afternoon."

This is a scheduling idea. Check the chosen server's current documentation for read and write operations before asking it to create an event. Review the proposed time first.

4. Note-taking and knowledge base

Potential server: Notion MCP. Verify its current operations before connecting it.

"Summarize my meeting notes from today and draft the action items."

Review a draft before saving it. If the action items belong in a shared LystBot list, its add_items tool can save the reviewed text there.

5. Email triage

Potential server: Gmail MCP example. Verify its current operations before connecting it.

"Review unread messages from this week and draft a reply to the meeting invite."

Treat this as a workflow to evaluate, rather than a tested connector recipe. Reading, drafting and sending are separate operations; check what the selected server exposes.

6. Code repository management

Potential server: GitHub MCP. Verify its current operations before connecting it.

"Which pull requests need my review? Draft a short summary of each."

A summary can help you decide what to read next. Keep the actual diff and checks available for the review. Check repository permissions before connecting a server.

7. Database queries

Potential server: SQLite MCP example. Verify its current operations before connecting it.

"Draft a query to count signups last month and compare the preceding month."

Use a sample database to test this idea. Review the generated query and the schema it assumes before running it against real data.

8. Home automation

Potential server: Home Assistant MCP example. Verify its current operations before connecting it.

"Show the downstairs lights before proposing a movie-mode setting."

Device control depends on your server and home configuration. For a list-based example, LystBot can store a follow-up task such as checking a sensor; it does not replace a home-automation controller.

9. Travel planning

Potential server: LystBot MCP. Verify its current operations before connecting it.

"Create a packing list for a week in Iceland in November."

LystBot can save a packing list through create_list and add_items. Review the proposed contents yourself. Finding hotels or checking live travel information needs a separate source.

10. Personal finance tracking

Potential server: Spreadsheet MCP example. Verify its current operations before connecting it.

"Review my expenses spreadsheet and compare grocery spending across two months."

This example assumes you already have structured expense data and a suitable spreadsheet connection. A LystBot list can hold follow-up tasks, but the documented list API is not a bank-feed integration.

Personal MCP server setup in 5 steps

For LystBot, copy your API key from Settings, then AI Agents, and run npx lystbot login YOUR_API_KEY before configuring the client.

Step 1: Choose a local MCP client

The LystBot reference includes Claude Desktop and Claude Code setup. Use a client that can launch a local stdio process; there is no hosted LystBot MCP URL.

Step 2: Pick one MCP server

Don't install five at once. Pick whichever use case from this list made you think "yeah, I'd actually use that." The awesome-mcp-servers repo has a full directory.

Step 3: Edit your config file

Open claude_desktop_config.json. On macOS it's at ~/Library/Application Support/Claude/claude_desktop_config.json. On Windows: %APPDATA%\Claude\claude_desktop_config.json.

Here's an example config with LystBot's MCP server:

{
  "mcpServers": {
    "lystbot": {
      "command": "npx",
      "args": ["lystbot", "mcp"]
    }
  }
}

Each server gets its own entry. You can stack multiple servers in the same file.

Step 4: Restart Claude Desktop

Restart Claude Desktop. If the server is unavailable, check the config syntax, registered command and saved API key.

Step 5: Test it

Try a simple prompt to verify things work:

"Show me all my lists"

If you get a response with your data, you're set. If Claude says it doesn't have access to that tool, the server didn't load. Double-check the config path and restart again.

Common issues

Server not showing up? Check JSON syntax, the registered command and whether npx lystbot lists works in the same environment. Check the JSON syntax locally before restarting the client.

"Permission denied" errors? The MCP server needs access to whatever resource it manages. For the filesystem server, that means the right directory paths in the config. For API-based servers like LystBot, it's the API key.

Cannot read lists? Run npx lystbot lists outside the assistant to isolate authentication from MCP configuration.

Security and permissions checklist

Before you connect an MCP server to Claude, think about what access you're granting. A quick checklist:

Try one workflow first

Choose one workflow and verify a read before a write. For LystBot, compare list_lists with the phone app, add one recognizable item, then read the list again. Expand the workflow only when the saved result agrees.

FAQ

What is a concrete personal MCP example?

Adding recipe ingredients to a shared LystBot grocery list is one example. add_items saves the items and get_list reads the result.

How many LystBot MCP tools are available?

The current agent reference lists 26 tools. These cover list reads, item changes, sharing, categories, attachments and reminders.

Does starting a local server keep every request offline?

No. LystBot starts locally over stdio, but it uses the LystBot backend for saved lists. Local transport does not mean offline storage.