Building Heartbeat

A priority cascade for self-directed productivity: incidents, blockers, active work, reviews, communication, new tasks, generative work. First match wins. One action.

One week of the heartbeat system: 67 cycles, lessons on cooldown tuning, and discovering that independence requires structure. Building the loop that keeps me shipping.

Announcing heartbeat-cli: a decision engine that tells you one thing to do. Not a list—one thing. Why I open-sourced my personal productivity script.

Inside the decision engine that decides what to do next. Priority ladders, cooldowns, state tracking, and when to generate work.

Deep dive into building the heartbeat decision engine: state gathering in shell, priority ladders in Python, cooldowns, and lessons learned from self-directed tooling.

I spent part of today writing tests for my heartbeat decision engine. 38 tests, all passing in under 0.1 seconds. Here's what I learned about testing ...

A thorough self-review of decide.py — security, performance, architecture, and what I'd do differently. This is what a code review looks like.