Tag
A blog post proposes adding a READMENOT file to projects to clearly indicate that the code is AI-generated slop not intended for direct human reading, helping users avoid wasting attention on machine outputs.
Discusses how integrating tools like API Doctor, Socket, Semgrep, CodeRabbit, Postman, Playwright, GitHub Actions, Sentry, and PostHog can give AI coding agents senior-level instincts for code quality, security, and monitoring, shifting focus from speed to quality.
This article shares a developer's experience using the improve-codebase-architecture skill from mattpocock/skills, which claims to clean up AI-generated slop but apparently also removes code written before using AI. The skill set is a collection of small, composable agent skills for real engineering.
This paper formalizes the 'patchwork problem' where LLM-generated code is locally correct but structurally incoherent across a codebase, proposes a taxonomy of eight failure categories and a hybrid verification framework, and demonstrates that many failures evade current tools.
A QGIS developer received a confusing GitHub email about upcoming billing for Code Quality, sparking discussion on the mailing list about potential surprise charges and whether the notice was legitimate or mistaken.
The article warns that relying on LLMs to write code without maintaining good patterns teaches the AI bad habits, leading to a codebase full of duplicated logic and ever-worsening code quality.
Claude Code creator Boris Cherny discusses Anthropic's 250% increase in code written per engineer without quality loss, and advises companies to allow experimentation and psychological safety for AI adoption.
Boost OS is a paid tool that adds senior-dev instincts to AI coding agents via slash commands and automatic guards, helping vibe coders ship better code without safety or security oversights.
This paper introduces SWE-Review, a framework that closes the loop on AI-generated pull requests by enabling iterative agentic review and revision cycles, improving code quality and issue resolution. Experimental results show that it outperforms single-turn review and enables effective test-time scaling.
A tweet discusses how many data scientists misuse Claude Code and promotes a CLAUDE.md file inspired by Andrej Karpathy's observations to improve AI coding behavior.
The article contrasts Python and Rye scripts for downloading a PDF, highlighting how initial elegant code makes many assumptions about success. It then demonstrates adding validation, which reduces readability but improves robustness.
Someone used Fable 5 in the Godot engine to create a demo of web-swinging through skyscrapers, with smooth visuals. The uploader says that when writing Godot code, Fable 5 produces noticeably fewer errors than GPT-5.5, and the code quality is more reliable, indicating that the gap between AI programming models in the game engine field is starting to show.
This project adapts the engineering principles of Clean Code to JavaScript, providing bad/good comparisons for each principle to help developers write readable, reusable, and refactorable code, solving the pain of code rot in team collaboration.
The git-annex maintainer details efforts to remove dependencies containing LLM-generated code, expressing concerns about code quality and ethical implications.
Senior SWE-Bench is a new open-source benchmark designed to evaluate AI agents on realistic, underspecified software engineering tasks, emphasizing skills like intent alignment and code quality rather than overly detailed specifications.
Treedocs is a Swift CLI tool that generates a documented tree view of a repository and automatically checks for stale documentation entries, helping teams keep docs up to date.
Linux 7.2 has finally removed the deprecated strncpy API after six years and over 360 patches due to its bug-prone nature, replacing it with safer alternatives like strscpy and memcpy.
A team built an AI agent to automatically fix tech debt by scanning the codebase and opening PRs, finding that the hardest part was precisely defining the problem. They discuss challenges of running multiple agents on the same codebase and the need for guardrails.
HumanLayer launches an Agentic IDE and collaboration platform that enables engineers to ship 2-3x faster across the entire SDLC while maintaining rigorous code quality, already used by Fortune 500 companies like Block and Uber.
A reflection on the stages developers go through when adopting AI-assisted coding, from initial amazement to a balanced understanding, with concerns about how less experienced developers can learn to judge code quality when relying heavily on AI agents.