<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Computer on BigOrangeQWQ Blog</title><link>https://blog.orangeqwq.top/en/categories/computer/</link><description>Recent content in Computer on BigOrangeQWQ Blog</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Tue, 14 Apr 2026 00:08:01 +0800</lastBuildDate><atom:link href="https://blog.orangeqwq.top/en/categories/computer/index.xml" rel="self" type="application/rss+xml"/><item><title>Prefix-Cache-Friendly SubAgent Derivation: A fork()-Style Design</title><link>https://blog.orangeqwq.top/en/posts/2026-04-14-llm-ideas-fork/</link><pubDate>Tue, 14 Apr 2026 00:08:01 +0800</pubDate><guid>https://blog.orangeqwq.top/en/posts/2026-04-14-llm-ideas-fork/</guid><description>This post records some of my personal ideas about LLM agents.
Recent Work Recently, I built a CLI tool for my own daily use (OMG CLI).
While designing it, I borrowed heavily from kimi-cli.
When it comes to CLI design, many people in the industry (including me) usually think about two things:
How to save context. How to keep attention focused. So today I want to share my different perspective and practice on these two points.</description></item></channel></rss>