LystBot

AI agent task list: shared groceries and todos

An AI agent task list in LystBot stores the tasks your assistant creates in the same app you use on your phone. Run npx lystbot login YOUR_API_KEY, then connect through claude mcp add lystbot -- npx lystbot mcp, or use the CLI directly.

Route Command or endpoint Use
MCP npx lystbot mcp Interpret a meal plan or conversation
CLI npx lystbot lists Read lists from a shell
REST GET /api/v1/lists Read lists in an automation

See the LystBot agent setup for the full reference. Related guides: Claude shopping list setup, HTTP automation for grocery lists.

AI chat adding items to a list app

AI agent list management for groceries

Grocery lists are the most obvious starting point because they happen while you're doing something else. You're cooking and realize you're out of cumin. You're reading a recipe and need eight ingredients. You're chatting with your AI about meal plans for the week.

Here's what an ai grocery list looks like with Claude Desktop and LystBot's MCP server:

You: "I want to make shakshuka this weekend. Add what I need to my grocery list."

Claude: "I've added these items to your Grocery list: canned tomatoes, bell pepper, onion, garlic, eggs, cumin, paprika, and feta cheese."

LystBot list detail showing grocery items added by AI Push notification when items are added to a shared list

Your partner gets a push notification. The list is on both phones. Nobody had to open the app and type anything.

The same thing works through the API. If you're building a recipe parser or a meal planning tool:

curl -X POST https://lystbot.com/api/v1/lists/LIST_ID/items \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"id": "NEW_ITEM_UUID", "text": "Flour", "quantity": 1, "unit": "500g"}'

REST adds one item per request, with a UUID v4 id and a text field. Replace NEW_ITEM_UUID with a new UUID before sending. For several ingredients, the CLI accepts comma-separated text and MCP provides add_items. We wrote a full walkthrough of the grocery use case with Claude: Using Claude as your MCP-powered grocery assistant.

Download LystBot free on the App Store and Google Play and try this yourself. The MCP server setup takes about two minutes.

AI task management from conversations

Todo lists are trickier than groceries. Groceries are simple: items go on, items get checked off at the store. Todos have priorities, deadlines, and context. This is where letting AI manage lists earns its keep.

Say you just finished a meeting. You tell Claude:

You: "That meeting had three action items. Add 'send revised proposal to Sarah by Friday,' 'research competitor pricing,' and 'schedule follow-up for next Thursday' to my work todo list."

Claude: "Done. I've added all three to your Work Todo list."

Or through the CLI, which works well inside coding agents and terminal workflows:

npx lystbot add "Work Todo" "Send revised proposal to Sarah by Friday"
npx lystbot add "Work Todo" "Research competitor pricing"
npx lystbot add "Work Todo" "Schedule follow-up for next Thursday"

The items sync to the app on your phone. You can check them off from anywhere. If you share the list with a colleague using a share code, they see the same items in real time.

What makes an ai todo list better than typing items yourself? Context. You can say "I need to prep for the board meeting next week" and let the AI break that into specific tasks. It remembers what you discussed earlier in the conversation. It can batch-add items instead of making you think of each one separately.

Connect your agent in 5 steps

Use the following steps to connect Claude Desktop to the local MCP server.

Step 1: Download LystBot

Get it on your phone from the App Store or Google Play. Create a list or two so you have something to work with.

Step 2: Copy your API key

Open Settings, then AI Agents in LystBot, and copy the API key.

LystBot settings screen showing API key

Step 3: Open your Claude Desktop config

Find claude_desktop_config.json:

Step 4: Add the LystBot MCP server

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

Before starting MCP, authenticate with npx lystbot login YOUR_API_KEY using the key from step 2.

Step 5: Restart Claude Desktop and test

Restart Claude Desktop, then test a read request:

"Show me all my lists"

If Claude returns your lists, you're set.

Common issues: If Claude says it can't find the tool, check the JSON for syntax errors (a missing comma breaks everything silently). If npx times out, your network might block npm registries. Check that npx lystbot mcp can start in the same environment. A 401 means authentication failed; check the saved key.

Three ways to connect, one backend. The MCP protocol spec is open, and LystBot runs as a local stdio server. The agent reference includes configuration for Claude Desktop, Claude Code and Cursor. The REST API handles automations. The CLI handles terminal workflows. All three sync to the same lists on your phone.

The Sunday review: recurring automation

A scheduled script can create a fresh list and add the weekly staples. The example below uses the CLI so it handles record IDs and item parsing.

For example, schedule a script on Sundays at 8am to create a dated grocery list and add staples:

#!/bin/bash
set -e
# Log in once with npx lystbot login YOUR_API_KEY before scheduling.
LIST_NAME="Groceries $(date +'%b %d')"
npx lystbot create "$LIST_NAME" --type shopping
npx lystbot add "$LIST_NAME" "Milk, Eggs, Bread"

Set up with crontab -e and add 0 8 * * 0 /path/to/sunday-groceries.sh. You wake up, your grocery list is already started.

Then during the week, whenever you're chatting with Claude about meals or recipes, you say "add this to my grocery list" and the items pile up alongside the staples. By Saturday you have a complete list without ever having opened the app to type anything.

The same pattern can create a Monday task list with recurring items such as reviewing the budget or backing up photos. Test the schedule and read the result before leaving it unattended.

The combination of scheduled automation (cron + API) and ad-hoc additions (AI agent + MCP) covers both the predictable and the spontaneous. That's what makes ai agent list management stick. For more automation ideas using the REST API, see our post on grocery list API automations.

Get started

LystBot is free during launch. Its CLI, MCP server and REST API use the same API key.

Download LystBot free on the App Store and Google Play

Set up the MCP server using the docs, tell Claude to add something to a list, and see it appear on your phone. Once the workflow clicks, you won't go back to typing items by hand.

FAQ

Where does the agent save a task?

It saves the task in a LystBot list through MCP, REST or the CLI. The mobile app reads the same list.

Should a fixed schedule use AI?

A fixed schedule can run a CLI script or an HTTP request directly. Use the assistant when the input needs interpretation, such as turning meeting notes into proposed tasks.

What if the agent chooses the wrong list?

Stop further changes and read the available lists again. Specify the intended list and verify one small change in the app before expanding the request.