Original Summary

Been working on a quantitative trading platform for 3~ years, at first 2.5 years off and on, then full-time for the last eight months when agents became capable. I started using agents seriously after writing 30K lines of code and found an agentic workflow I could trust, and I turned that workflow into Pullboard. So, this holds up in real systems.<p>Essential Loop:<p>claim -&gt; build -&gt; verify (different agent reviews, see website.)<p>Primitive Concepts:<p><a href="https:&#x2F;&#x2F;pullboard.dev&#x2F;tutorials&#x2F;" rel="nofollow">https:&#x2F;&#x2F;pullboard.dev&#x2F;tutorials&#x2F;</a><p>What makes this workflow valuable is the &quot;Agentic Institutional Knowledge&quot; built up over time outside the codebase in the Pullboard history; e.g., items, shouts and doctrine all get referenced, discussed and those lessons remain available to future agents instead of disappearing into chats. The best part is I hardly re-explain myself.<p>You do have to steer agents and make decisions, but this is quite nearly autonomous development. Anyways it is early and open source. Feedback welcome. Thank you!!

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/10/10 02:46:41