Microsoft's open-source tools were hacked with the objective of stealing passwords from AI developers. This incident highlights software supply chain vulnerabilities and the growing threat to professionals working with artificial intelligence.
The article introduces ForgeAI Pipeline Intelligence, an open-source Jenkins plugin that utilizes 8 specialized AI analyzers to detect architectural drift and supply-chain risks within CI/CD pipelines. It provides a release verdict (SHIP_IT, CAUTION, HOLD, or BLOCK) to enhance code quality and minimize post-pull request issues.
This official Anthropic repository serves as a directory for high-quality Claude Code Plugins, designed to enhance its capabilities and foster open-source AI innovation. It invites developers to explore and contribute to a collaborative community focused on practical AI applications.
This article analyzes OpenClaw, an open-source AI agent framework for personal use that integrates with messaging apps. The author expresses mixed feelings, finding it exciting yet hesitant to recommend it broadly, prompting a deeper investigation into the project.
This guide explains how to integrate Promptfoo, an open-source CLI tool for LLM testing, with CometAPI, a unified OpenAI-compatible API. This allows developers to test over 500 models from a single key, often at a lower cost than direct providers.
The author introduces Web Speed, an open-source tool that parses HTML web pages into easily readable sitemaps for AI agents, making them faster and cheaper. The project includes a global cache of sitemaps to further speed up agents, currently accessible only via the paid API version.
Qwen3.6-27B, a new dense, open-source model, has been released, boasting flagship-level agentic coding power that surpasses its predecessor, Qwen3.5-397B-A17B. It also features strong reasoning across text and multimodal tasks, supports thinking/non-thinking modes, and is available under the Apache 2.0 license.
Syll is an open-source, self-hosted multimodal agent harness that unifies various interfaces (APIs, CLI, GUI) for personal automation. It enables users to teach procedures and inspect agent execution through multimodal evidence, supporting modularity and auditability.
The open-source community is endorsing OpenEnv for agentic Reinforcement Learning development. This initiative highlights collaborative efforts in advancing AI.
Chaperone-Thinking-LQ-1.0, a 4-bit GPTQ + QLoRA fine-tuned DeepSeek-R1-32B model, has been open-sourced. It achieves 84% accuracy on MedQA, close to GPT-4o, while being only ~20GB in size and 1.6x faster than the base model.
Slopper GitHub Action is designed to combat low-quality AI-generated contributions in open-source projects. Its purpose is to maintain code quality and integrity by detecting and rejecting "AI Slop".
The Heretic Free Software Project received a legal notice from Meta Platforms, Inc. concerning derivatives of Meta's "Llama" AI language models. In response, Heretic has removed these materials from its model weight repositories, sarcastically expressing gratitude for the opportunity to align with the "agenda of the global corporate oligarchy".
The author found reading non-fiction with a tree-structured AI agent effective but inaccessible for many users. To solve this, they built pi-books, an open-source, local-first reading companion that transforms books into navigable, tree-like conversations, addressing the limitations of linear AI chat threads.
A new book and open-source code are released, detailing how to build modern LLM architectures like GPT-2, Llama 3, and DeepSeek from scratch in PyTorch. It explains the architectural changes required to transform GPT-2 into Llama 3 and implements DeepSeek's advanced features.
This post introduces the Nova Sonic Test Harness, an open-source framework designed for scalable evaluation and rapid iteration of Amazon Nova Sonic voice agents. It automates multi-turn conversations, uses LLM-as-judge techniques to validate quality, and can detect audio hallucinations without requiring a microphone.
The author shares their journey of creating THOR, a website offering guidance for open-source contributions, inspired by a personal experience. The text introduces project setup using Node.js, Playwright, and Passmark, emphasizing the importance of securing API keys, including one for OpenRouter.
Cloudflare released Unweight, a lossless compression system that reduces LLM size by 15-22% without sacrificing output accuracy. The tool, which saves roughly 3 GB of VRAM on Nvidia H100 GPUs for Llama-3.1-8B, has been open-sourced on GitHub with plans to extend compression.
LIDARLearn is a unified PyTorch library for 3D point cloud deep learning, supporting 56 ready-to-use configurations and built-in cross-validation. It also automates publication-ready LaTeX PDF generation after training, making it ideal for researchers in 3D computer vision and remote sensing.
![We’re proud to open-source LIDARLearn [R] [D] [P]](/cdn-cgi/image/width=3840,quality=75,format=webp/https://preview.redd.it/53o0rt8wfxvg1.png?width=640&crop=smart&auto=webp&s=17dbca74f07ea72d1d32060979cecfdc47ab51fe)
OpenSkynet is a project showcased on Hacker News, introducing an "AI Terminator." It refers to a GitHub repository for an artificial intelligence-related initiative.
Deepseek has announced the release of DeepEP V2 and TileKernels. These projects, available on GitHub, represent new developments in their AI-related software offerings.