Tag
A Lobsters discussion asking whether teams allow LLMs to write both production code and tests, with the author considering a policy restricting AI generation to only one of the two to avoid tautological, over-engineered results.
An opinion piece arguing that Common Lisp is now the best programming language because LLMs shift the bottleneck from writing code to fast feedback loops, live image-based development, interactive debuggers, and macro-driven domain-specific languages that cut code volume and token costs.
Colin Armstrong reflects on how LLMs changed the economics of personal software, detailing how he built heade.rs — a polished domain lookup tool showing DNS, registration, latency, and email config — with Claude Code in about an hour and deployed it to Cloudflare for less than the cost of a coffee a month.
A researcher reports spending three months running Claude Code and Codex on Erdős Problem #993 (unimodality of independent-set sequences in trees), only to learn that Tong Zhang and Wei Li posted a full proof last week, along with two Lean 4 formalizations that claim a clean build. The result is not yet peer-reviewed.
A discussion post arguing that if LLMs write most code, we should design a single AI-oriented language where users describe intent and constraints while the compiler decides implementation details across languages like Go, Java, or SQL.
Garry Tan revealed he deleted about a thousand lines of markdown from gstack because frontier models are now good enough that the old prompt-engineering tricks and heavy instructions are no longer necessary.
MAGS introduces a multi-agent framework that uses formal verification with Dafny to generate executable programs with safety guarantees from LLM coding agents, achieving 100% success in producing verified code across domains like CUDA kernels and robotic tasks.
The article argues that software engineers must differentiate themselves from AI models by focusing on areas like deep codebase familiarity and technical communication to maintain their value in the industry.
The article presents the Agentic Coding Index, a method to compute the intelligence density of LLMs by aggregating various coding benchmarks and normalizing for parameter count.
The article shares a method using an agent.md file to define coding style preferences for LLM-assisted development, improving code quality by reducing repetitive feedback.
WaveScope is an MCP server that applies wavelet transforms to codebases, providing LLMs with multi-resolution structural context to improve code understanding and editing, addressing context rot and structural awareness.
FrontierSmith automatically generates diverse open-ended coding problems from closed-ended tasks, improving LLM coding performance on benchmarks through enhanced agent interactions and training data synthesis.
本文深入解析了FactoryAI的Missions多智能体架构，通过角色分工、验证合约与结构化交接机制，实现了可在生产环境中连续稳定运行数十天的自动化编码系统。该设计将软件工程瓶颈从人工执行转向人类注意力管理，为开发者提供了可落地的长期多智能体协作方案。
An open-source maintainer explains why they now prefer LLM-generated code over community PRs, arguing that AI assistance reduces risk and friction while shifting the value of contributions toward feedback, bug reports, and design discussions.
A single CLAUDE.md file became GitHub's top-trending repo with 44k weekly stars by distilling Andrej Karpathy's LLM coding advice into four principles.