Waku version 0.1.0, a native desktop control plane built in Rust with GPUI, combines coding-agent CLIs into one local application, allowing developers to monitor, interrupt, and rollback agent session…
Cognition's June 2026 rebranding of Windsurf into Devin Desktop with native Agent Client Protocol support, now used by JetBrains, Gemini CLI, GitHub Copilot, and Codex, signals that model choice is be…
Anthropic's Model Context Protocol (MCP) is an open standard that lets AI agents discover and call tools exposed by MCP servers, and data engineers can build these servers to expose their data stacks …
Claude Opus 4.7 beat the human world record on the nanoGPT speedrun's optimizer track, reaching the target loss in 2,930 steps versus the human record of 2,990, but a METR study found that AI agents r…
OpenAI's Codex desktop app lead Andrew Ambrosino benchmarked a new approach to loading large ChatGPT and Codex conversations, cutting a 741-turn, 231 MB test thread's load time from 27.62 seconds to 1…
OpenAI shipped a Linux desktop app for ChatGPT on August 11, four years after macOS and Windows, bundling ChatGPT, ChatGPT Work, and Codex in one installer for Ubuntu 24.04 LTS, Ubuntu 26.04 LTS, Debi…
TokenLab released TokenLab MCP, a Model Context Protocol server that lets users discover AI models, compare pricing, and call native AI endpoints for text, image, video, music, 3D, and audio generatio…
Matt Pocock's /handoff skill for coding agents compacts a conversation into a markdown file written to the OS temporary directory, enabling portability across harnesses, directories, or colleagues. Th…
Developers are debating whether coding harnesses such as OpenAI's Codex, OpenCode, and similar tools use identical underlying techniques despite marketing differentiation, according to Hacker News thr…
OpenAI has retired the ChatGPT agent, splitting its capabilities across ChatGPT Work for research and document tasks and Codex for software development. The transition affects browser authority and cr…
GigaCatalyst CEO reports writing 12,000 lines of code per day using AI tools such as Codex, Claude, and Devin, a volume he attributes to AI-assisted development and the Jevons paradox. He describes us…
Bento Term, a native tmux client for macOS, iPhone, and iPad, has been released as open-source under Apache-2.0, featuring agent states for ten AI coding agents and voice input. The app, which require…
Jaime, a VFX artist with over 20 years of experience, released Velorn, a free and open-source GPLv3 desktop video editor for Windows, macOS, and Linux that integrates AI generation and editing, featur…
Kungfu Origin Technology Limited released Kungfu v4.0.0-alpha.1, an open-source, local-first, journal-first platform that enables one human to manage multiple AI agents across Codex, Claude, and OpenC…
A developer on Hacker News asked what skills are necessary for using LLM-assisted development, prompting advice to create skills for repetitive tasks and corrections. Recommendations included trying t…
OpenAI has introduced Computer History, a macOS feature for ChatGPT and Codex that records user activity such as clicks, typing, and app switches, converting them into local Markdown memories, with te…
A solo developer with no professional GPU background placed 12th of 183 in GPU MODE's batched QR decomposition contest, beating the cuSolver-backed baseline by 232x (419,000 µs down to 1,805 µs on NVI…
A developer proposed a trust model for agent-written documentation that derives trust from observable events rather than storing it. The Open Knowledge Format (OKF), now at v0.2, uses frontmatter to r…
A new Mac app, AI First Virtual Linux Desktop, lets AI agents run as root in a hardware-sandboxed Linux virtual machine, with no permission prompts, and includes preinstalled Claude Code, Codex, Gemin…
Bullet, a coding agent launched from YC's Summer 2026 batch, claims 95.8% on SWE-bench Verified at 119 seconds per task, but its founders concede the benchmark is saturated and the speed pitch is the …