Full-text search across 35672 articles. Combine with topic and date filters; results sorted by relevance.
Claude Code versions 2.1.154 through 2.1.158 contain a regression that corrupts the streaming delivery of tool results back to the model, causing results to arrive empty-then-late in bursts, duplicate inline blocks up to…
The Zig Software Foundation has banned all AI-assisted contributions to the Zig programming language codebase, prohibiting code generated, paraphrased, edited, brainstormed, or debugged by large language models. Zig pres…
Two models have tied the all-time record on agent coding benchmarks, scoring 90% with zero hard fails. Qwen3 Coder 30B A3B achieved the score in 28 seconds at $0.0004, while DeepSeek Chat (original) reached 90% in 59 sec…
Weekly crypto code commits have fallen roughly 75% since early 2025, dropping from about 850,000 to 210,000, while active developers declined 56% to around 4,600, according to data from analytics platform Artemis. The co…
The article describes the creation of The Polyglot Protocol, an open-source framework designed to address common failure modes in AI coding agents, such as ignoring repository conventions, inventing APIs, and skipping va…
Google Workspace offers a 14% discount on plans for the first three months, available through promo codes valid until the end of 2026. The suite includes three main plans—Starter, Standard, and Plus—which vary in storage…
The article compares three local LLM-based code completion tools—Zed AI, Continue, and Cursor—highlighting their strengths and weaknesses for different workflows. It notes that local models offer privacy, no subscription…
According to the article, "vibe coding" – the practice of using AI tools to create apps without traditional programming skills – is expanding from desktop to mobile devices. Google announced updates at its I/O conference…
Researchers have developed Speculative KV coding, a lossless compression method that reduces the size of large language model KV caches by up to approximately 4x using a smaller, faster predictor model. The technique, in…
A new AI workflow system called "The 8 Systems" claims to replace a five-person engineering team by stacking multiple Claude Code hacks into automated, end-to-end production workflows. The system, which costs $200 to $50…
Developer Daniel Lublinsky released Skill Atlas, a free, open-source graph-search plugin for Claude Code that adds a 'searchable' tier between enabled and disabled skills, cutting per-session token usage from 2,544 to 33…
A developer has built NexusMem, a local-first memory engine for AI coding agents that indexes git history, shell commands, project docs, and assistant transcripts into a SQLite database, serving ranked context over MCP o…
The Yale Budget Lab, led by executive director Martha Gimbel, argues that Washington should fix existing tax code gaps before drafting new taxes on AI-generated wealth, warning that current disparities between labor and …
Blacksmith raised $45 million in a Series B round led by Peak XV Partners, valuing the San Francisco startup at $550 million as AI coding tools increase the volume of software requiring validation. The company, which pro…
Browser Agent, developed by Pierre Barreau, achieved an 88% success rate on the BU Bench, outperforming Browser Code's 78%, while also reducing cost to $5.37 from $8.34 and cutting execution time to 32,694 seconds from 4…
Anthropic's Claude Code v2.1.212 on macOS in IntelliJ IDEA terminal sent a user's real email address as the User-Agent string in a curl command without explicit confirmation, according to GitHub issue #78431 filed by a u…
Anthropic announced that auto mode will be turned on by default in Claude Code for Pro, Max, and Team accounts starting Aug. 14, allowing the AI agent to proceed without human prompts unless actions are destructive or ir…
A developer measured the fixed token overhead of spawning subagents in Claude Code and found it to be roughly 436,000 tokens per agent before any useful work, dwarfing the payload cost. In a comparison of a review pipeli…
Databricks Cost Optimizer, a read-only FinOps toolkit and Agent Skill for Codex and Claude Code, identifies cost spikes and applies only approved optimizations, achieving a 68.5% reduction in daily effective-list-price u…
Meta Superintelligence Labs released Muse Glimmer, a 30-billion-parameter open-weights model optimized for local agent workflows, under an Apache 2.0 license. The model runs on a single consumer GPU and targets local cod…