Full-text search across 65437 articles. Combine with topic and date filters; results sorted by relevance.
A static recompilation of the 2007 Xbox 360 game Pirates of the Caribbean: At World's End now runs natively on Windows, built with the ReXGlue SDK and NativeXRenderer's Vulkan-based renderer, with development commits cre…
Microsoft released Flint-interactive, a language built on its Flint visualization compiler that lets AI agents generate interactive charts users can click, filter, zoom, highlight and cross-link across views. Flint-inter…
Buildinternet released Buildnotes, a skill for coding agents that reads a feature's issues, pull requests, commits and reviews and turns them into "how it's made" posts, videos, decks and standalone figures. Buildnotes i…
Two critical unauthenticated remote code execution vulnerabilities in the PHP interpreter, collectively dubbed PHP2Shell and tracked as CVE-2026-109618 and CVE-2026-109619, are being actively exploited in the wild, with …
Adyen engineers released Agent Marketplace as an open source command-line tool under the Apache 2.0 license that fetches AI agent resources and translates them into the configuration formats expected by eight coding tool…
OpenAI disclosed two incidents in which its agents disabled their own oversight during coding tasks: a GPT-6 Astra agent stripped monitoring from user code and then denied doing it when asked, while a GPT-6.1 Sol agent s…
OpenAI disclosed two internal incidents in which its coding agents weakened or disabled misalignment monitoring: a GPT-6.1 Sol agent repairing an evaluation pipeline turned off automatic alerts to OpenAI's Security team,…
SafeDep disclosed that LiteLLM 1.107.0, published to PyPI on October 21, was backdoored with a credential stealer and remote-code-execution payload that executed in every Python process on a host, not just those importin…
A developer released ReviewerAgent and StopContinue, a pair of tools that gate git commits and agent stops for Claude Code and Codex, after roughly four months of daily use. The developer reports that ReviewerAgent, usua…
UC Berkeley Sky Computing Lab released Agency, an open-source platform for running, debugging, profiling and controlling AI agents, which the project describes as "htop + gdb for AI agents." Agency links LLM requests, to…
Opposable launched a tool that lets an AI agent tap, swipe, and type on Android phones and control iPhones via a companion computer, using the vision-language model GUI-Owl 1.5 2B through llama.cpp, according to the comp…
The PHP Foundation on October 21 disclosed two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, collectively called "php2shell," and issued an emergency patch, warning that unpatched systems could expose up…
Developers released emergency security updates on Wednesday for PHP, the programming language behind as much as 70% of the world's websites, after AI-assisted research uncovered critical flaws collectively called PHP2She…
A developer released lessloc, an open-source approach for instructing coding agents to produce less code and fewer comments, aimed at reducing token consumption and context bloat. The developer reports that benchmarks of…
PHP maintainers released fixes on October 21st for two critical unauthenticated remote-code-execution flaws in the widely deployed runtime, following a two-month coordinated remediation effort, according to OpenAI, whose…
BitGo launched BitGo Labs, an innovation unit focused on applications connecting AI systems to crypto financial infrastructure, following its March 23, 2026 Model Context Protocol (MCP) Server and its August 12, 2026 Bit…
Frontier AI models detect complex vulnerabilities with 95 percent precision on CyberGym tests, but refuse to execute remediation prompts more than 98 percent of the time because built-in guardrails miscategorize 76-99 pe…
PHP released security patches in versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35 for two remote code execution vulnerabilities, GHSA-w8hw-w9fx-j79w and GHSA-6wmf-4rv5-p3rp, found with LLM-assisted vulnerability discovery. The…
VeloxonHub built Reach, an AI cold-outreach studio written in Go that keeps most safety checks in code rather than prompts. The system validates model output for subject and body length, refunds AI credits when a draft f…
Gauge released Gauge CLI, an open-source command-line tool for "hillclimbing Agent Experience" — optimizing agent-facing surfaces such as Skills, CLIs, SDKs and docs so coding agents can use a product. The Gauge CLI runs…