#vibe-coding

Google's white paper explores "vibe coding" and agentic engineering as new software development paradigms, proposing an AI-driven SDLC where agents handle coding while humans focus on direction and validation.

Vibe coding—relying on AI to generate code without understanding it—is analogous to a system design interview, where the key skill is evaluating trade-offs and guiding the AI toward correct solutions rather than writing code from scratch. The article argues that this approach shifts the developer's role from implementation to architecture and oversight, making it a valid modern engineering skill.

A developer using AI tools like Claude and Cursor has created over 80 daily video games. The latest release, World Domination, is a simplified clone of the board game Risk. The creator shares the project and invites feedback.

Cory Doctorow argues that understanding computers at multiple abstraction levels—from cardboard CARDiac to BASIC code to modern "vibe coding"—builds "legibility" and Fingerspitzengefühl essential for problem-solving, and that automated code tools serve as a funnel for cultivating deeper technical skills.

The Godot game engine has banned AI-generated and "vibe-coded" contributions, requiring contributors to certify their work is not produced by large language models. Violations may result in bans. The policy aims to address code quality and copyright concerns.

AI-generated websites often lack critical accessibility and legal compliance features, such as alt text, semantic HTML, and privacy notices. Developers using "vibe coding" tools risk lawsuits and fines under laws like the ADA, GDPR, and accessibility regulations if they deploy sites without proper auditing and legal safeguards.

A developer reflects on what they would do after achieving financial success from rapidly building and marketing numerous software products with AI assistance. They envision using AI to learn advanced mathematics, improve foreign language skills, and explore hardware or robotics. The author also complains about HN's AI comment detection system falsely flagging their manually written replies.

A developer created a systems language called Tig using AI-driven "vibe coding," but the project stalled after exhausting free tokens. Turning away from AI dependence, they successfully built a lexer for a calculator in raw C by using traditional resources like man pages and YouTube tutorials, concluding that learning to code independently, though slower and harder, is more rewarding.

Cory Doctorow argues that AI coding paradoxes resolve when distinguishing "today's task" code from "accretive work." Drawing on the concept of "canonization," he explains that while personal, disposable code has value, the AI industry's focus on reverse centaurs consumes software canon without replenishing it.

The post humorously captures the typical "vibe coding" experience, where developers rely on AI tools to generate code but often struggle with debugging, understanding the output, or maintaining control over the project's direction.

The author argues that "vibe coding" (using AI to generate code without understanding) is ending soon as AI shifts to structured reasoning. Those relying on vague instructions will be left behind, and developers must learn fundamentals now to avoid obsolescence.

The article describes a day-75 project in which the author uses Claude (an AI assistant) to "vibe-code" a video game called "Corner the Market," likely exploring the process of building a game through conversational AI interaction rather than traditional coding methods.

The article asks which platform—Fable 5 by Anthropic or Replit—users would prefer for building a "vibe coded" application, seeking comparisons between the two.

In a podcast discussion, DHH talks about the development of Basecamp 5, the concept of "vibe coding" (an AI-assisted approach where developers describe intent rather than writing code line-by-line), and the future direction of the Rails framework in an era of increasing AI integration in software development.

Google's whitepaper redefines the traditional Software Development Life Cycle (SDLC) for the agentic era, introducing "Vibe Coding" where AI agents handle coding, testing, and deployment. The paper presents a new framework that shifts developer roles from writing code to orchestrating AI-driven development workflows and managing system outputs.

The blog introduces a three-phase workflow for software development using Claude Code: Vibe Coding (rapid prototyping with natural language), Agentic Engineering (structured, test-driven agentic loops), and Human Review (code review and refinement). It emphasizes moving from chaotic AI-generated code to disciplined, production-ready engineering practices.

The developer created persist-os, a local CLI tool that records architecture decisions and standards in plain repo files and uses a "persist doctor" command to check if AI-generated code contradicts those rules, skips obvious steps, or lacks tests. It runs locally without a network and retains rules across sessions.

The author describes a month spent "vibe-coding" at extremely slow pace (0.01x velocity), reflecting on the experience of building software with AI assistance when deliberately working far below normal speed. The piece critiques the rushed culture of AI development and explores what can be learned from slowing down the coding process to a crawl.

Fredrin.dev presents a Kanban tool designed for vibe coding workflows within Claude Code and Codex environments.

A Twitter post accuses someone of stealing code from open-source and enterprise-licensed sources rather than creating it themselves, coining the term "vibe code" to describe the alleged theft.

The article explores how Claude Code enables a shift from "vibe coding" (prompt-driven, loosely guided development) to agentic engineering, where AI assists in structured, autonomous software development tasks like debugging, refactoring, and testing through natural language commands and multi-step reasoning.

A hobby project website offers a per-project AI use disclaimer template under CC0 license. The creator argues that open source maintainers have a duty to disclose their extent of AI use, aiming to replace vague "vibe coded" accusations with transparent, responsible tool use disclosures.

The article argues that GitHub is increasingly filled with low-quality AI-generated code, creating a "vibe coding crisis" where quantity overshadows quality. This flood of auto-generated repositories makes it harder for developers to find reliable, human-written code and threatens the platform's value as a trusted code resource.

"Vibe coding" with AI produces unmaintainable, undocumented code from the start, creating legacy code before it ships. The author warns this leads to technical debt, hidden bugs, and systems difficult to debug, contrasting it with traditional engineering practices focused on clarity and long-term maintainability.

Using Claude Code, the author ported the Moebius 0.2B image inpainting model from PyTorch/CUDA to ONNX WebGPU, enabling it to run entirely in a browser. The AI agent handled the entire conversion, weight publishing, and UI without manual coding. The final demo works in Chrome, Firefox, and Safari with caching support.

Godot has clarified its stance on AI in game development, stating it permits limited AI assistance while rejecting the label of being "vibe coded," as the engine continues to address misconceptions about its development practices.

ExperiOps is a platform and service focused on the maintenance and ongoing management of "vibe coded" projects, helping developers sustain and operate software built through intuitive or experimental coding approaches.

The author used AI-assisted "vibe-coding" to quickly build several niche Mac apps for personal needs, including a text splitter, karaoke timer, metronome calculator, and photo scanning tool. The approach enables rapid prototyping without deep coding expertise, producing functional but unpolished results.

Building an app is harder due to assembling services (DevOps, payments, auth) than writing code. The author envisions an AI agent that can deploy an entire app from a single command, handling docs, API keys, debugging, and production deployment without human clicks or web pages.

The article discusses how "vibe coding" (using AI to generate code from natural language) is creating a new SDLC where developers focus more on specification, review, and architecture rather than manual implementation. It argues this shifts the bottleneck from writing code to defining intent and validating outputs.