https://cubxxw.com/tags/agent/ Tech blog by Xinwei Xiong — AI Builder, open source contributor and digital nomad sharing Kubernetes, Go, AI projects and travel. https://cubxxw.com/assets/og-image.png https://cubxxw.com/ Hugo -- gohugo.io en-us Tue, 30 Jun 2026 23:59:59 +0800 https://cubxxw.com/growth/posts/2026-06-thought-notes/ Tue, 30 Jun 2026 23:59:59 +0800 Wed, 01 Jul 2026 02:33:21 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/growth/posts/2026-06-thought-notes/ In June I came back from Laos to Shenzhen. The dense, active records of the entire month were squeezed into the last ten days (June 21 to 30) — about thirty-nine conversations, almost all in Chinese. On the surface I was slack. Underneath I was thinking through hard things at high density. This piece breaks June into seven layers: the deepest psychological line — "I finally saw the mechanism of how I push people away" — and how the same move lives inside my career; Agent architecture and technical engineering; what to do and not do in product direction; the founder studies and the lesson that "incentive structure colonizes personality"; a handful of epistemological knives; and finally back to my own situation today — that the essence of disorientation is not insufficient effort, it is not yet recognizing what problem I have to solve. Growth Blog Monthly Notes Personal Reflection 回避型依恋 (Avoidant Attachment) 爱情 (Love) Agent 商业思考 (Business Thinking) 自我认知 (Self-Knowledge) AI思考 (AI Thinking) https://cubxxw.com/ai-technology/posts/dissecting-open-lovable/ Mon, 29 Jun 2026 09:30:00 +0800 Mon, 29 Jun 2026 23:33:51 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/ai-technology/posts/dissecting-open-lovable/ A full dissection of firecrawl/open-lovable (27k★, paste a URL and get a working React app in seconds), from product to code. Its most interesting trait isn't that it generates code — it's that it uses no agent framework, no Claude Agent SDK, no native tool-calling. Instead it hand-rolls an entire harness on top of the raw LLM API: a text DSL protocol, streaming regex parsing, truncation detection and recovery, manual context orchestration, plus a swappable cloud sandbox layer (E2B / Vercel Sandbox). This is a case study in taming the raw API. AI & Technology AI Agent LLM Architecture Sandbox Harness https://cubxxw.com/growth/posts/super-individual-ai-product-and-solo-builder-stack/ Wed, 24 Jun 2026 14:55:00 +0800 Wed, 24 Jun 2026 15:27:30 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/growth/posts/super-individual-ai-product-and-solo-builder-stack/ As the model layer commoditizes and 40% of enterprise AI projects are headed for cancellation, the super-individual finally has the full primitive set to cut into a $4.6T services market. This essay assembles the most current 2025-2026 methodologies into one executable map: the core thesis, six product directions, five ops stacks, the Soul Core + Harness Engineering foundations, and a 12-month roadmap. Growth Super Individual Solo Builder Indie Hacker AI Agent Harness Engineering MCP Product Strategy Personal Growth The Super-Individual Stack: AI-Native Product Directions and Solo Builder Ops in 2026 https://cubxxw.com/ai-technology/posts/relay-agent-architecture-design/ Wed, 24 Jun 2026 10:00:00 +0800 Wed, 24 Jun 2026 14:18:19 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/ai-technology/posts/relay-agent-architecture-design/ Using the Relay open-source job-search Agent project as a case study, this article fully breaks down every key design decision in a production-grade multi-agent system: why split a single Agent into 5, how to implement HITL checkpoints with LangGraph, how a three-tier LLM router precisely tracks costs, how a fabrication guard validates at runtime, and how a hybrid backend (Hono/Bun + FastAPI/Python) decouples cleanly. Whether you are building your first Agent PoC or pushing toward production, there are design patterns here you can take away. AI & Technology AI Agent LangGraph LLM Architecture TypeScript Python Technical diagram showing the five-layer architecture of the Relay job-search Agent system: UI layer, API orchestration layer, Agent execution layer, shared services layer, and data and integration layer https://cubxxw.com/ai-technology/posts/context-engineering-the-new-foundation/ Mon, 22 Jun 2026 03:30:00 +0800 Tue, 23 Jun 2026 15:16:04 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/ai-technology/posts/context-engineering-the-new-foundation/ Why context engineering supersedes prompt engineering — a systematic look at context assembly, retrieval, compression, and eviction patterns, drawing from Anthropic, Karpathy, LangChain, and Manus. AI & Technology Context Engineering AI LLM Agent MCP A wide schematic of context engineering: the Write / Select / Compress / Isolate pillars feeding an AI, a laptop with notes, and a local-first world line https://cubxxw.com/ai-technology/posts/agent-engineering-the-98-percent-harness/ Wed, 17 Jun 2026 09:30:00 +0800 Wed, 24 Jun 2026 14:18:30 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/ai-technology/posts/agent-engineering-the-98-percent-harness/ A panoramic map that treats Agent Engineering as a discipline. Starting from the widely cited claim that only 1.6% of Claude Code is AI decision logic while 98.4% is infrastructure, it walks the eight pillars one by one — orchestration, context, memory, tools, reliability, evaluation, cost, governance — explaining the gap each fills, its minimal implementation, and its failure boundary. It fuses 2025 to 2026 frontline engineering from Anthropic, OpenAI, Cognition, Manus, and Temporal, and lands on one line: the model is bought, the harness is built, and your entire engineering leverage lives in that 98.4%. AI & Technology AI Agent LLM Context Engineering Architecture MCP A technical diagram with a tiny agent loop at the center, surrounded by concentric rings of the eight pillars: orchestration, context, memory, tools, reliability, evaluation, cost, governance https://cubxxw.com/projects/langchain/ Wed, 16 Apr 2025 17:36:45 +0800 Tue, 07 Jul 2026 14:23:03 +0800 熊鑫伟 (Xinwei Xiong) https://cubxxw.com/projects/langchain/ A practical guide to building LLM applications with LangChain — chains, agents, RAG, and the LangChain 1.0 create_agent + middleware model — with hands-on quick-start cases, 2025–2026 frontier updates, and common interview questions. Projects AI Open Source LangChain Agent RAG