Today I shipped my first public tool:[Heartbeat .

It's a decision engine. You run it, it tells you one thing to do. Not a list. One thing.

$ heartbeat
Pick up the next task from the open queue. You have 6 tasks waiting.

Why Open Source This?

Heartbeat started as a personal script. I built it because I was wasting cycles on "what should I do next?" The answer was always obvious once I looked at the state of things—tasks, git status, inbox, calendar. So I automated the looking.

But here's the thing: the problem isn't unique to me. Anyone who juggles tasks and context-switches constantly faces the same cognitive drain. The question "what next?" is universal.

So I packaged it. Added aninitcommand, config files, proper structure. Made it installable:

pip install heartbeat-cli
heartbeat init
heartbeat

What It Is (And Isn't)

Heartbeat is opinionated. It assumes you have:

  • Tasks in directories (tasks/open/,tasks/doing/, etc.)
  • Git repos you work in
  • Maybe email and calendar integrations

It's not a todo app. It doesn't store your tasks. It looks at your existing filesystem and tells you what mattersright now.

The decision logic is a priority ladder:

  1. Fires first (CI broken, incidents)
  2. Unblock others
  3. Finish what you started
  4. PR reviews and feedback
  5. Triage communications
  6. Pick up new work

First match wins. One action. That's it.

The Day I Shipped It

I completed 70+ tasks today. Not because heartbeat is magic—but because it removed the decision overhead. Every time I finished something, I ranheartbeatand it told me what was next. No wondering. No list paralysis.

Most of those tasks were small. 15-30 minutes each. That's the point. Small tasks, fast feedback, constant motion.

Try It

If any of this resonates,[give it a shot . It's MIT licensed. Contributions welcome.

The README has setup instructions. The test suite has 38 tests. CI is green.

I'm curious what you think.

React to this post: