Vibe Coding
Post-Code AI EraThe curated reading list on vibe coding — Andrej Karpathy's term for the post-code AI era. 19 essential pieces covering definition, tools (Claude Code / Cursor / Codex), the agentic engineering boundary, and real case studies.
“Vibe coding is Karpathy's term for the post-code AI era — when coding agents cross the threshold from tool to collaborator, and the bottleneck moves from typing to judgment. This vault is the curated reading list for that shift.”
About this vault
Vibe coding is Andrej Karpathy's term for letting an AI agent write code by feel rather than by spec. In 2026 the tools (Claude Code, Cursor, Codex) have crossed the threshold where 80%+ of a working codebase can be generated — and the bottleneck moves from typing to judgment. This vault is the curated reading list for that shift: the definition debate (is vibe coding an anti-pattern?), the boundary with agentic engineering (two frames of the same shift), 2026 tools comparison (why Codex-users love Claude and vice versa), the post-code AI workflow Karpathy predicted, and real-world case studies — from SwiftUI apps to viral apps that went down. 19 essays spanning Karpathy's blog, Simon Willison, Boris Cherny (Claude Code creator), and the builders actually shipping in the vibe-coding era. Each piece has an AI summary so you can decide what to read in seconds.
19 articles
All 19 articles
GitHub - langchain-ai/open-swe: An Open-Source Asynchronous Coding Agent
Open-source async coding agent by LangChain — run SWE tasks in background, review diffs when ready. CI-style automation for AI-assisted development.
The End of Coding: Andrej Karpathy on Agents, AutoResearch, and the Loopy Era of AI
Karpathy on the end of traditional coding: agents handle execution, autoresearch handles discovery, and we enter the 'loopy era' where AI iterates on its own output.
When Your Vibe Coded App Goes Viral—And Then Goes Down
What happens when your vibe-coded app hits the front page and crashes. A cautionary tale about skipping infra fundamentals when AI writes your code.
Head of Claude Code: What happens after coding is solved | Boris Cherny
Boris Cherny (Head of Claude Code) on what comes after coding is solved: orchestration, taste, and knowing what to build become the scarce skills.
跟朋友探讨，AI 时代，如何考察候选人，一致的结论是，需要现场看候选人的 vibe coding 能力。 以前考察古法编码技能，会针对算法、数据结构、解题速度、编码规范等做细致了解，也考察专业知识。 但这些技能，随着 Coding Ag…
How to evaluate developer candidates in the AI era — vibe coding changes the hiring bar. Traditional coding interviews miss what actually matters now.
1 year of vibe coding — retrospective
vibe coding 这个词是洗澡时随手发的推文，结果变成了行业术语。一年后回看：预测不了什么会火，但方向是对的——自然语言正在取代编程语言。
GitHub - langchain-ai/open-swe: An Open-Source Asynchronous Coding Agent
Open-source async coding agent by LangChain — run SWE tasks in background, review diffs when ready. CI-style automation for AI-assisted development.
Optimal LLM-assisted coding — diversifying workflows
编程工作流在分化，不是一个工具吃天下：~75% 用 Claude Code 做复杂任务，剩下靠 autocomplete、chat、手动编辑各司其职。关键是根据任务性质切换工具。
Boris Cherny, the creator of Claude Code, shared his entire setup. He runs 5-10 Claudes in parallel. Half his coding ha…
Boris Cherny's full Claude Code setup revealed: config, workflows, and prompting patterns from the person who built the tool.
10x Your Coding Agent Productivity
10 practical tactics to multiply coding agent output: better prompts, parallel agents, structured context, and knowing when to intervene vs let it run.
Google Stitch + Claude Code，DESIGN.md让Vibe Design更有设计感
Google Stitch + Claude Code with DESIGN.md makes vibe design feel designed. A design spec file bridges AI generation and visual quality.
为什么程序员钟爱Codex，Vibe Coding的人却离不开Claude？一文弄懂它！
Why devs love Codex but vibe coders stick with Claude — different mental models, different strengths. Codex for precise edits, Claude for open-ended creation.
Claude coding notes — 80% agent coding, 20% edits
几周之内从 80% 手写代码 / 20% agent 变成 80% agent / 20% 手动修改。这不是渐进式变化——是突变。Claude Code 是第一个让他相信 agent 真能用的工具。
Programming changed dramatically — coding agents crossed threshold
2025 年 12 月是编程的分水岭：coding agent 在这之前基本不能用，之后突然能用了。不是渐进式进步，是相变——模型质量、长程连贯性、韧性同时跨过了阈值。
A great resource if you want to understand how AI coding agents work. Learn Claude Code walks through building a minimal…
A solid resource for understanding how AI coding agents work under the hood — architecture, tool use, and the loop that makes them effective.
Vibe coding SwiftUI apps is a lot of fun
Simon vibe codes macOS apps with Claude Opus 4.6 and GPT-5.4 — a full SwiftUI app fits in a single text file, no Xcode required. He builds Bandwidther (network monitor) and Gpuer (GPU monitor) as menu bar tools in one sitting.
Vibe coding MenuGen
第一个完整的 vibe coding 实战案例：纯自然语言 prompt 构建出可用的 app。代码变得免费、临时、可丢弃——用完一次就扔也不心疼。
Can coding agents relicense open source through a "clean room" implementation of code?
The chardet library debate: can coding agents pull the Compaq BIOS clean-room trick in hours instead of months? Simon on the legal and ethical gray zone of AI-generated 'clean room' reimplementations that sidestep copyleft licenses.
🧑🚀 Day 6 of the @cursor_ai #vibejam Proudly sponsored by @cursor_ai + @boltdotnew + @heyglif Prizes to win (submit …
Day 6 of Cursor VibJam: real-time progress log showing what works and breaks when building with AI in a competitive jam setting.
Start reading, not hoarding.
Import this vault to Burn 451 and actually read what matters.
Content attributed to original authors. Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.