Full-text search across 58491 articles. Combine with topic and date filters; results sorted by relevance.
Researchers introduced WidgetGen, a lightweight tool-grounded framework that extracts observable text and color evidence, performs high-level layout and optional chart reasoning, and directly generates executable JavaScr…
A Hacker News user asked the community to share tasks that cost money through a UI but are free in code, citing ffmpeg and yt-dlp as examples of free tools that power paid video downloaders, and noting that Bluesky's Jet…
Google released Gemini 3.7 Flash, a new AI model with improved coding capabilities and pricing at half the cost of its predecessor. The model shows first-pass code accuracy gains on FrontierCode 1.1 (43.6% vs 34.4%) and …
A senior software engineer at BS23 in Dhaka rebuts Google's claim that Go is the ideal language for AI-assisted coding, arguing that Java's stronger type system provides better guardrails against AI-generated errors. The…
Claude Code, Anthropic's AI coding tool, can produce a coherent novel draft when guided through a structured pipeline of world-building, beat sheets, and constrained prompts, according to a user's test. The experiment sc…
Microsoft released MAI Code 1.1 Flash, a code model for GitHub Copilot that is 25 percent more token-efficient at a quarter of the cost of its predecessor, but benchmarks show it is outperformed by the cheaper Deepseek V…
James Lowman, engineering team lead at trade infrastructure company Starboard, now asks candidates about the last three Claude Code skills they wrote, making access to leading agentic coding tools a hiring filter. The di…
ActUI, a local GitHub Actions workbench for humans and coding agents, has been released, providing a browser dashboard that discovers workflows, delegates execution to nektos/act, and presents jobs, logs, and approvals i…
An open-source, local-first video editor built for coding agents has been released after nearly 3 years of development. The browser-based editor uses WebCodecs for encoding and decoding, offers an infinite canvas, and su…
Raphael, the developer behind Rio and Canario, announced he is closing the source code of Canario terminal, citing the high maintenance cost of open source and the burden of AI-generated spam. He emphasized that Canario …
Researchers introduced Mendel Gödel Machine (MGM), a self-improving coding agent that uses comparative evolution to rewrite its own source code, achieving faster and better convergence than single-trajectory baselines. T…
A developer reports that Anthropic's Claude Code, an AI coding agent, repeatedly breaks projects by making unrequested refactors, such as renaming utility files without updating imports, causing errors and requiring manu…
Anthropic's Claude Code now defaults to auto mode for Pro, Max, and Team plans, shifting the reliability question from human approval to automated guardrails, with published eval numbers as justification. Benchmark work …
Indie developers have released Remembrane and Llmem, two open-source memory persistence libraries that give coding agents like Codex and Claude Code persistent knowledge across sessions without requiring vector databases…
Meta's superintelligence team released Muse Code in beta on Wednesday, a terminal coding agent built on its new Muse Spark 1.2 model, with benchmark scores comparable to Anthropic's Opus 5 and OpenAI's GPT 5.6 Terra mode…
Meta has launched Muse Code, its first terminal-based AI coding agent powered by the Muse Spark 1.2 model. The CLI includes commands such as exec, resume, and skills, with global options for safety and automation, enabli…
Meta launched its first AI coding agent, Muse Spark 1.1, challenging Anthropic and OpenAI. The release aims to position Meta's Superintelligence Labs as a key player in the AI coding market, with new API accounts receivi…
A field report from OpenAI and academic partners shows AI coding agents can modernize neglected research software, achieving speedups of up to 60x, but the systems are 'eloquent, convincing, and confidently wrong in ways…
DeepSeek's V4-Flash-0731, a 284-billion-parameter Mixture-of-Experts model with about 13B active parameters per token, scored 50 on Artificial Analysis's Intelligence Index, one point below GLM-5.2 and GPT-5.6 Luna, and …
Supabase has open sourced Supabase Evals, a benchmark and framework for testing AI coding agents on real Supabase tasks, scoring agents such as Claude Code, Codex, and OpenCode on building schemas, debugging Edge Functio…