<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0">
    <channel>
      <title>LLM Wiki</title>
      <link>https://kingqiu.github.io/LLMWiki</link>
      <description>最近的10条笔记 on LLM Wiki</description>
      <generator>Quartz -- quartz.jzhao.xyz</generator>
      <item>
    <title>Agent Observability</title>
    <link>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/observability</link>
    <guid>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/observability</guid>
    <description><![CDATA[ Agent Observability Definition Agent observability is the practice of monitoring, tracing, and evaluating LLM agents in production by capturing their reasoning chains, tool invocations, and decision points—enabling debugging, performance optimization, and quality assurance for non-deterministic agen... ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Agent Planning</title>
    <link>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/planning</link>
    <guid>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/planning</guid>
    <description><![CDATA[ Agent Planning Definition Agent planning is the capability of an LLM agent to decompose complex goals into executable sub-tasks, reason about dependencies and sequencing, and dynamically adjust strategies based on intermediate outcomes. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>RAG (Retrieval-Augmented Generation)</title>
    <link>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/rag</link>
    <guid>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/rag</guid>
    <description><![CDATA[ RAG (Retrieval-Augmented Generation) Definition RAG is a read-only retrieval mechanism that grounds language models in external, static knowledge bases (documents, wikis, manuals) by searching for semantically similar content and injecting it into the prompt context at inference time. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Agent Reflection</title>
    <link>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/reflection</link>
    <guid>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/concepts/reflection</guid>
    <description><![CDATA[ Agent Reflection Definition Reflection is a metacognitive strategy where an agent evaluates, critiques, and iteratively improves its own outputs. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>RAG vs Agent Memory: Architectural Boundaries</title>
    <link>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/synthesis/rag-vs-memory-boundary</link>
    <guid>https://kingqiu.github.io/LLMWiki/ai-agent-architecture/synthesis/rag-vs-memory-boundary</guid>
    <description><![CDATA[ RAG vs Agent Memory: Architectural Boundaries Analysis A common architectural mistake in agent systems is treating RAG and Agent Memory as interchangeable—or worse, attempting to use RAG as a substitute for memory by dumping conversation logs into a vector database. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>China Agent Landscape</title>
    <link>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/china-agent-landscape</link>
    <guid>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/china-agent-landscape</guid>
    <description><![CDATA[ China Agent Landscape Definition 定义 The China agent landscape refers to the ecosystem of AI agent platforms, vendors, and deployment patterns specific to the Chinese market, characterized by platform integration (WeChat, DingTalk), government support, and private deployment requirements. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Enterprise CLI Design</title>
    <link>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/enterprise-cli-design</link>
    <guid>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/enterprise-cli-design</guid>
    <description><![CDATA[ Enterprise CLI Design Definition 定义 Enterprise CLI design refers to the architectural patterns and principles for building command-line interfaces that are optimized for AI agent consumption while meeting enterprise requirements for security, compliance, and integration. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>High-Privilege Agent</title>
    <link>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/high-privilege-agent</link>
    <guid>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/high-privilege-agent</guid>
    <description><![CDATA[ High-Privilege Agent Definition 定义 A high-privilege agent is an AI agent with access to production systems, databases, and APIs that can make consequential changes. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Institutional Intelligence</title>
    <link>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/institutional-intelligence</link>
    <guid>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/institutional-intelligence</guid>
    <description><![CDATA[ Institutional Intelligence Definition 定义 Institutional intelligence refers to AI systems designed for organizational coordination and decision-making, as opposed to individual productivity. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>On-Premise Infrastructure</title>
    <link>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/on-prem-infra</link>
    <guid>https://kingqiu.github.io/LLMWiki/enterprise-agent-china/concepts/on-prem-infra</guid>
    <description><![CDATA[ On-Premise Infrastructure Definition 定义 On-premise infrastructure refers to the hardware, software, and networking components deployed within an enterprise’s own data centers for running AI agents, as opposed to using public cloud services. ]]></description>
    <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
  </item>
    </channel>
  </rss>