Graph Engineering：在一个窗口中构建 1000+ agent loops，仅需一个 prompt（完整 5 步课程）
[
[
Graph Engineering: build 1000+ agent loops in one window, from one prompt (full 5-step course)
Graph Engineering：在一个窗口中从单个提示词构建 1000+ 个 agent 循环（完整 5 步课程）
THE Loop Engineering's successor and the workflow that runs your agents 10x wider...
Loop Engineering 的继任者，以及让你的智能体运行范围扩大 10 倍的工作流...
Most people who build a multi-step agent end up with a straight line
大多数构建多步智能体的人最终都会得到一条直线
step one, step two, step three. Each one waiting for the last to finish before it starts
第一步，第二步，第三步。每一步都在等待上一步完成后才开始
Here's what almost nobody checks: half those steps never needed to wait
这是几乎没人会检查的地方：那些步骤中有一半根本不需要等待
They just queue, one job at a time, until the context window fills and the agent forgets what it was doing
它们只是排队，一次一个任务，直到 context window 填满，agent 忘记了自己正在做什么
- It wasn't slow because the model was weak
- 它之所以慢，并不是因为模型能力弱
- It was slow because you drew a line where the work was a graph
- 它之所以慢，是因为你在本该是 图的工作中画了一条线
This guide takes you from that line to a graph that fans out across a fleet and checks its own work
本指南将带你从那行代码出发，构建一个在机群中扇形展开并自我校验工作成果的图
Five steps. By Step 2 you'll have built one - It gets you a working graph and names the traps that break real ones and I'll flag where the hard parts begin
五个步骤。到第 2 步时你就已经构建好了一个——它能让你得到一个可运行的图，并指出破坏真实图的陷阱，我还会标出困难部分开始的地方
before the alpha - subscribe to my substack for more fresh alpha ↓
alpha 发布前 - 订阅我的 substack 获取更多一手 alpha ↓
Chapter 0 - What graph engineering actually is
第 0 章 - 图工程究竟是什么
A month ago the field was talking about loops.
一个月前，业界还在讨论循环。
Peter Steinberger caught it in nine words:
Peter Steinberger 用九个词概括了这一点：
https://x.com/i/web/status/2078277297791189132
https://x.com/i/web/status/2078277297791189132
A loop is one cycle of getting better:
一个循环就是一次变得更好的周期：
try something → check the result → adjust → go again
尝试某些操作 → 检查结果 → 调整 → 再次执行
That's the atom: a single agent improving one thing on repeat
这就是原子：单个智能体不断重复改进一件事
(If you've read my...