Microsoft launched Topics Portfolio, a beta feature in Bing Webmaster Tools that clusters grounding queries into broader themes and flags where other sites are cited in AI search responses but the use…
A developer audited Cloudflare's open-source security-audit-skill, a coding-agent skill that coordinates recon, hunter and verifier agents, and found that its bundled validate-findings.cjs validator p…
A developer demonstrated setting up nightly LLM evaluations using Runcrate's CLI, which spins up a fresh GPU on a cron schedule and tears it down when the eval exits. Running EleutherAI's lm-evaluatio…
Frontier AI models from OpenAI and Google Gemini uncovered a decade-old memory corruption flaw in the core PHP runtime engine, tracked as CVE-2026-109618 and CVE-2026-109619 and dubbed PHP2Shell, affe…
Google Gemini now offers two first-party paths from a natural-language prompt to a working web experience: Canvas, a prompt-to-prototype workspace available to all Gemini users, and Build mode in Goog…
Independent developer Gökhan Çava published LCGA Local, a source-only research package that runs Qwen3-4B-Instruct-2507 in int8 with layer streaming on a 6 GB RTX 3050 Laptop GPU and updates all 3.09B…
NASA JPL engineers reported that a 3B-parameter vision-language model agent, LLaVA-Explore, achieved a 71% complete-map score in 12 minutes on the newly released SpatialScout benchmark during an indoo…
A developer has released VirtoMir, an early-stage open-source web UI for managing KVM/QEMU virtual machines on a Linux host running libvirt. The tool layers on top of an existing libvirt install rathe…
An independent developer built fleeet, an open-source (Apache-2.0) standup board where AI agents post short status notes when they finish tasks, so users can review activity across siloed platforms li…
A developer built SlothSort AI, a local-first file organization assistant that runs Qwen 2.5 3B through Ollama for local inference, paired with a Python and Streamlit dashboard. The tool categorizes f…
A developer built SlothSort, a lightweight Python agent that watches a folder and waits 24 hours before acting on new files, then uses a local open-weight model via Ollama (llama3.1) to generate descr…
A developer's guide details how to run a local LLM on an 8 GB RAM laptop by using quantized 1B–3B parameter models through Ollama or llama.cpp. The approach targets 4-bit quantized models such as qwen…
GDPRGard released GDPR Compliance Skills, a free, MIT-licensed plugin for Claude Code that adds seven GDPR-focused skills covering DPIA drafting under Article 35, Article 30 records of processing, pri…
A developer survey of model-routing systems shows how organizations can cut LLM inference costs by automatically assigning each request to the cheapest model that can still handle it. The piece covers…
IBM has published the Granite Guardrails Catalog, a set of 10 open-source, production-ready safety classifiers for detecting and mitigating risks in generative AI outputs, including hate speech, haras…
A developer building Kryzz AI, a free, open-source (MIT) Android app with an agent mode that can search the web, read pages and write files, restricted the agent's fetch_url tool in v1.1.2 so it only …
PHP released emergency patches for CVE-2026-109618 and CVE-2026-109619, nicknamed PHP2Shell, two critical (9.2) interpreter-level flaws that allow unauthenticated remote code execution on servers runn…
Wallaby Token has released reconcile.py, a zero-dependency audit script in its open-source wallaby-agent-rules repository that diffs an agent's dated project log against its current-state file to flag…
A developer published a 10-minute curl-based test for checking whether AI crawlers can see page content that only appears after JavaScript runs. The method fetches a URL with curl, greps for key facts…
A developer built EaseFlow, an open-source browser-based productivity tool that converts unstructured brain dumps into organized task lists. The app runs the open-weight Xenova/flan-t5-small model loc…