<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Autonomous-Coding on korchasa@*ops</title><link>https://korchasa.dev/tags/autonomous-coding/</link><description>Recent content in Autonomous-Coding on korchasa@*ops</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 08 May 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://korchasa.dev/tags/autonomous-coding/index.xml" rel="self" type="application/rss+xml"/><item><title>lumatale: a team of agents builds the service almost without me</title><link>https://korchasa.dev/posts/2026_05_08_lumatale_autonomous_agents/</link><pubDate>Fri, 08 May 2026 00:00:00 +0000</pubDate><guid>https://korchasa.dev/posts/2026_05_08_lumatale_autonomous_agents/</guid><description>&lt;p&gt;I wanted to test &lt;a href="https://github.com/korchasa/flowai-workflow" rel="nofollow noopener noreferrer external"&gt;flowai-workflow&lt;/a&gt; on something more or less complex and real, since the usual &amp;ldquo;home&amp;rdquo; tasks are too simple, and work tasks are mostly built around ci/cd pipelines.&lt;/p&gt;
&lt;p&gt;In the end I decided to make a fully autonomous and cheap process for building services. The service is developed by a group of agents based on flowai-workflow. It plays all the roles itself, including product manager. I sometimes act as hitl, but not too often. Product and architect — sonnet 4.6 from the claude code subscription. Developer and QA — glm 4.6 from the annual z.ai subscription. Right now development speed is bottlenecked by glm limits, bugs, and my own meddling in the sdlc.&lt;/p&gt;</description></item></channel></rss>