agentscmdSee how it works

The to-do list your agents can read.

Plan your day like you always have. Hand any task to Claude Code, Cursor or any MCP agent, and review what comes back.

FREE DURING BETA · SIGN-IN OPENS SHORTLY · MCP + REST API
WORKS WITH →CLAUDE CODECURSORWINDSURFGITHUB COPILOTZEDANY MCP CLIENT

Less copy-paste.
More done.

Your agents see the same list you work from, and write what they did straight back into it.

01

Speak it, it lands

Tell your agent, or type it yourself. Either way the task lands scheduled and prioritized, and the quick-add shows its parse before it commits.

02

A list that sorts itself

Inbox, Today, Upcoming, Anytime. Anything overdue rolls into Today on its own, so nothing gets lost in a folder.

03

Clears in seconds

Dense, calm, keyboard-first. j and k to move, x to complete, c to capture. Fly through the list without touching the mouse.

04

One goal a day

Write down the one thing that would make today a win. In the evening your agent reads the day and scores it honestly.

05

One key per agent

Every agent connects with its own scoped, revocable token. See who added what; cut one off without touching the others.

06

MCP + REST underneath

A full MCP server and REST API, documented and scope-gated, so any agent or script can work your list.

HOW IT WORKS
  1. 1

    Connect your agent

    claude mcp add agentscmd --transport http <your MCP URL>

    Your personal URL and key come from the setup page, which opens with sign-in, and work in any MCP client.

  2. 2

    Hand off a task

    Ask your agent to plan your day or take something over. It reads the task and your list over MCP, gets to work, and writes what it did straight back.

  3. 3

    Review and check off

    What your agent finished shows up in your list, marked as its work. You review it, and you stay the one who closes things.

Questions,
answered.

What does "agent-first" mean?

Most todo apps are built for fingers: you type, you file, you drag. agentscmd is built so your AI agent can do that part over MCP, while you get a clean list and just check things off. The agent does the bookkeeping; you do the living.

Which agents work with it?

Claude Code, Cursor, Windsurf, Zed, GitHub Copilot and anything else that speaks MCP, including an agent you built yourself. You create one key per agent on the setup page; it takes about a minute.

Do I need an AI agent to use it?

No. agentscmd is a complete, fast todo app on its own: type "dentist friday 10am" and it lands scheduled. The agent is what makes it magical, not what makes it work.

Is my data private?

Yes. Your tasks are locked to your account, agent keys are stored hashed and can be revoked individually, and there is no AI on our side. Nothing reads your tasks except the agent you explicitly connected.

How do reminders work?

Honestly: there are no push notifications yet. Reminder times are saved, and your agent brings them up when you talk to it, for example in your morning plan. Notifications arrive with the phone app.

What does it cost?

Nothing during the beta. Pricing comes later, and beta users will hear about it first.

FREE DURING BETA · SIGN-IN OPENS SHORTLY

Give your agents a to-do list.