Загрузка видео...
Не удалось загрузить видео
SOMEONE ON THE CLAUDE CODE TEAM JUST EXPLAINED HOW TO MAKE A TASK RUN THE SAME WAY EVERY TIME Lydia Hallie uses dynamic workflows for any task she wants to rerun the same way. No need for hundreds of subagents: around 10 is enough for her. Two types of... show more
40,141 просмотров • 5 дней назад •via X (Twitter)
Комментарии: 12

a mí lo que me pasa es que luego no me acuerdo de que lo tengo guardado y lo vuelvo a pedir a mano

When a workflow pins which steps run and in what order, the payoff shows up at debugging time: reruns become comparable, so a drifting output can be diffed step by step instead of re-reading a whole transcript. 'It worked last week' becomes something you can investigate.

Explaining the same task every monday was my whole job for a while The rerun button is the real upgrade

"around 10 is enough" — enough for what exactly, nobody ever says what the 10 do

How does it handle error recovery if a subagent fails mid-run?

I keep a log of my agent's past mistakes and let my agent maintain it it every session.. making that a /workflow instead of a prompt is what I'll try next. does the file survive a big refactor?

yo soy la persona 1 y me acabo de dar cuenta ahora mismo

hahaha theres a lof of people like you

i just watched a bug report get a workflow and my improv career feels unstable

viene de Lydia Hallie, que trabaja ahí, así que me lo creo bastante más que a los hilos de siempre

wow, I didnt knew I was person 1 until now

rerunning a task the same way every time is what makes a workflow worth keeping. around 10 subagents also keeps it small enough that you can still read what each one did.

