get ai todo
The Mission Control for Coding Agents

A task list for you & your coding agents

Less managing.
More making.

The single, prioritized backlog built for autonomous coding agents.

Keep your projects in order. Give your AI agent a clear next task. Come back to a record of what changed.

Paid plans from $9.99/month. No free tier. 3 active MCP slots included.
Claude Code·Cursor·Antigravity·Codex·Cline
The next releaseExample project3 tasks MCP Connected
Your priorities, in orderSelect a task
01 — 03

A simple list. A deliberate order.

Task details

Remember the last open project

Ready for an agent

Open the workspace where I left it. Restore the selected project after signing in, on desktop and mobile.

Agent ReadyAny connected agent
Verification evidence:Awaiting agent claim via /backlog or /backlog-complete
Priority 01 · Available for claim
Your backlog, without the busywork.Illustrative preview · Click any task to inspect its agent lifecycle

Why Get AI Todo?

Stop copy-pasting backlogs.
Give agents an authoritative home.

Coding agents fail when backlogs are buried in 40-column kanban boards, Jira epics, or endless chat transcripts.

Generic Task Managers & Chat UI

Prompt fatigue & loose context

  • Manual copy-pasting of issue descriptions into chat prompts.
  • Complex kanban boards and story points that confuse AI context windows.
  • No direct tool access: agents cannot claim tasks or submit diffs.
  • Zero audit trail: did the agent actually run tests before closing?
Get AI Todo

Native MCP mission control

  • Agents query the backlog directly via Streamable HTTP Model Context Protocol.
  • Strict manual order: agents always execute priority 01 first.
  • Two-way supervisor loop: agents ask questions and submit diffs for review.
  • Remote kill switches: pause rogue or stale runs across any project in one click.

Engineered for autonomous coding

Everything agents need to work.
Everything you need to trust them.

Every feature is deliberately designed around the interaction between a developer and autonomous agents.

Native Model Context Protocol (MCP)

Connect Claude Code, Cursor, Antigravity, or Codex via Streamable HTTP. Run npx @getaitodo/cli login in your terminal to authenticate in seconds. Includes 3 active named token slots with instant rotation.

Streaming JSON-RPC 2.0 · OAuth & Token Auth

Autonomous Execution (/backlog-complete)

Your agent runs /backlog-complete to claim the top-priority task, launch specialized worker sub-agents, execute changes, run your local test suites, and stream progress milestones back to your phone.

One command · Local tool execution

Strict Manual Priority Order (01, 02, 03)

No hidden sorting algorithms, sprints, or estimations. Drag or use keyboard shortcuts to reorder. Priority 01 is always handled first. Bug, Feature, and Flag types keep intention razor-sharp.

Deterministic · Zero backlog confusion

Human-in-the-Loop Supervision

When an agent encounters ambiguity, it pauses in waiting_for_user so you can answer. When work finishes, it enters in_review with syntax-highlighted git diffs and test logs awaiting your approval.

Never unmonitored · Human holds final say

Remote Kill Switch & Leases

Pause agent execution across an entire project with one click. Tasks use lease epochs: if you pause a project or change requirements, running agents stop at their next check-in.

Instant project pause · Anti-runaway protection

Desk & Mobile Realtime Sync

A distraction-free 3-pane workbench on your desktop for deep work, paired with iOS and Android apps for mobile review. Get push notifications when an agent submits work while you are away.

Web workbench · iOS & Android push notifications

Connect your preferred environment

One command to connect.
Zero prompt pasting.

Get AI Todo connects directly to the tools you already use every day.

Claude Code CLIRun once in your project root terminal
claude mcp add get-ai-todo -- npx -y @getaitodo/mcp

From a thought to a finished task

A place for the work.
Not more work.

Get AI Todo holds the context. Your coding agent does the work in the tools you already use.

  1. 01

    Make the next step clear.

    Group tasks into projects, order them by priority, and add the details that matter. Use Bug, Feature, and Flag to describe the work.

  2. 02

    Connect your coding agent.

    Log in with the CLI or use an MCP token with Antigravity, Claude Code, Cursor, or another compatible tool. Your agent fetches project tasks without copying them into a prompt.

  3. 03

    Agent executes autonomously.

    Run /backlog-complete. The agent claims the top task, creates dedicated worker sub-agents, writes code, runs test suites, and streams real-time progress.

  4. 04

    See what was completed.

    Review the completion notes, the model used, and the completion time. Stay in control of what happens next.

The handoff, made simple

Your list. Your agent.
One shared context.

/backlogSee the tasks for your project in priority order.

/backlog-completeLet an orchestrator delegate tasks to sub-agents, run tests, and report back.

Install the provided skills in your supported agent client. Agents run in your coding environment, not on this website.

One small subscription

Everything you need.
Nothing to unlock.

Projects, tasks, completion details, and three active MCP tokens. Your subscription is tied to your account.

Monthly

$9.99/month

A little room to find your rhythm.
Get started

Yearly

Best value · Save 33%

$79.99/year

For a year of focused work ($6.67/month).
Get started

Also available at $2.99/week. Prices in USD, plus applicable tax. Subscriptions renew automatically; cancel future renewals in account settings. Extra active token slots are $0.99 each, one-time.

All core features in every plan.

3 active MCP token slots included.

Full access on Web, iOS, and Android.

Secure checkout powered by Dodo Payments.

A couple of useful details.

Does the website run my AI agents?

No. Get AI Todo is the shared task list. Your agent reads and updates it through MCP while running in your own coding client. Model subscriptions and usage are separate.

Can I use it without an AI agent?

Yes. Add projects, write task descriptions, set the order, and complete work yourself. Connect an agent when you want a hand.

How is agent access controlled?

Give each connection its own named token. Three active tokens are included, and you can revoke a token from the MCP page when it’s no longer needed.

What is the difference between /backlog and /backlog-complete?

/backlog allows your agent to read your ordered project backlog and understand immediate priorities. /backlog-complete commands the agent to autonomously claim the highest-priority open task, execute the code changes, run automated tests, and stream verification evidence back to you.

Does Get AI Todo store my repository code?

No. Your codebase stays entirely on your local machine. Get AI Todo only stores task titles, descriptions, project rankings, and the structured verification summaries and diffs returned by your agent when completing a task.

What happens if an agent gets stuck or makes an error?

Tasks claimed by agents have time-bounded concurrency leases. If an agent stops responding, the lease expires safely. Additionally, you can remotely pause agent access for any project in one click, immediately rejecting subsequent agent writes.