Cursor, the AI coding company, open-sourced Mixture-of-Kittens (MoK) on August 4th, a GPU kernel that it claims increased end-to-end training throughput for its Composer models by 41% against its prev…
The Model Context Protocol (MCP) 2026-07-28 spec, finalized on July 28, rewrites the protocol core from stateful to stateless, deprecates Sampling, Roots, and Logging with a 12-month grace period, and…
GitHub introduced stacked pull requests to help developers and coding agents break large AI-generated changes into smaller, reviewable layers. The feature addresses the challenge of reviewing massive …
Vega released Detection Skills on August 4, 2026, as an open standard for packaging cyber detection triage, investigation, and tuning logic into reusable agent workflows, with a public specification a…
A developer has built a workflow that runs every untrusted repository through a local LLM triage pass before opening it in an editor, after encountering a fake-recruiter take-home project that contain…
A production-focused playbook from the WildOctopus team outlines when to use transfer learning, full fine-tuning, LoRA, QLoRA, continued pretraining, and knowledge distillation for large language mode…
GitHub will retire GitHub Spark on github.com, ceasing new user sign-ups and app creation on August 4, 2026, with full shutdown on August 31, 2026. Existing users can export their apps before the shut…
A new browser-based lab at hego.red/jackpot offers ten deliberately broken LLM applications for hands-on security testing, running entirely client-side with no login or setup. The lab includes demos o…
NVIDIA released Alpamayo 2 Super, an open 34-billion-parameter reasoning vision-language-action model for autonomous vehicle development, combining a 32-billion-parameter Cosmos 3 Super Reasoner with …
Arize AI has launched Signal, a managed agent built into Arize AX that automatically reviews production traces on a recurring schedule, identifies recurring failure patterns, and turns them into prior…
Reware Labs released AI Code Security Cards, an open-source security knowledge layer that provides version-specific security guidance for AI coding agents, developed from Hossein Hajipour's PhD resear…
Netskope Threat Labs reported that cybercriminals are cloning popular GitHub repositories for AI tools and developer resources to distribute an infostealer, shifting from the ClickFix social engineeri…
Alibaba's Qwen3.8-Max model autonomously operated for 16 days on a public GitHub repository, opening and closing 151 issues, merging 127 pull requests, and landing 265 commits without human interventi…
Entangle, a new open-source tool from developer Gowtham Sai Yadav, lets users share live AI coding sessions from Claude Code, Codex, and opencode with a three-word code, with end-to-end encryption and…
Independent researcher Jahangir Nasimi is seeking an arXiv endorsement for his paper on AI-driven nuclear safety, presented at the IAEA International Conference on Topical Issues in Nuclear Installati…
GitHub has introduced code coverage automatic enablement in Code Quality settings, now in public preview for all GitHub Code Quality users on github.com. The feature uses AI to generate a coverage wor…
Av9n, a new third-party validation tool for AI agents, lets users check an agent's claims against external evidence such as GitHub CI results, addressing the need for independent evaluation of agent p…
On March 31, 2026, Chaofan Shou discovered that Anthropic's Claude Code source code was exposed via a sourcemap file in the npm package, revealing the entire source code of the AI coding CLI. The leak…
Pillar Security researchers disclosed on August 3, 2026, a novel attack against Google's open-source Agent Development Kit for Python, specifically targeting workflows in the gemini-cli GitHub reposit…
A developer built a network of AI agents to automate content creation for a Threads account, which grew to 32K followers, and used Linear for orchestration. To avoid per-seat costs, they developed a l…