AI Tools news and analysis on Web Pulse: 32796 curated articles tracking the latest AI Tools developments, tools, and research, updated continuously from vetted sources.
Google Cloud's Model Armor security service can be integrated with LangChain and the Agent Development Kit (ADK) for sanitizing prompts and responses in Generative AI applications. The second installment of the series de…
A new tool from imagesv2 removes white backgrounds from GPT Image 2 outputs, delivering transparent PNGs with preserved soft edges for e-commerce, stickers, and slide decks. The service processes user prompts and returns…
Claude Code skills are reusable instruction packs that any Claude Code session can call into when relevant, bundling instructions, optional resources, and triggering conditions. Users can write their own skills to captur…
A security researcher has discovered and exploited a 21-year-old use-after-free vulnerability in PHP's `unserialize()` function, affecting code paths that have been vulnerable since PHP 5.1 shipped in 2005. The bug, caus…
A developer has released a Python script that transplants extra tensors—such as Multi-Token Prediction (MTP) layers—from one GGUF file into another, enabling the creation of mixed-quantization models. The tool preserves …
A developer building a RAG (Retrieval-Augmented Generation) solution discovered that the most impactful feature was the chatbot's ability to say "I don't know." By integrating retrieval and tightening prompts, the chatbo…
A developer deployed the Qwen 3.6-35B-A3B FP8 mixture-of-experts model (3 billion active parameters) on a DGX Spark GB10 system using vLLM, achieving inference with a 262,144-token context window and FP8 KV cache. The co…
A free AI growth kit offering over $4,000 in startup credits, tactical GTM playbooks, and pre-built AI agent workflows is being widely shared among founders and operators. The resource, which includes partner credits fro…
OpenAI's GPT-5.5 outperformed Anthropic's Opus 4.7 and OpenAI's GPT-5.4 on 56 real coding tasks from two open-source repositories, passing more tests and producing patches that more often survived code review. Opus 4.7 w…
OpenAI has released Codex for Teams, a new feature that helps product teams manage multiple AI coding sessions by treating each session as a tracked unit of work. The tool addresses the growing challenge of coordinating …
Open source alternatives to Obsidian are emerging for developers and teams building AI workflows, with tools like Logseq, AppFlowy, VS Code, and Nimbalyst offering varying degrees of markdown portability and AI integrati…
A malicious npm package named "exiouss" was published on May 1, 2026, by the account "loltestpad" as a rebranded version of the previously removed "godsplan" package, now bundling a PowerShell script that decrypts and ex…
By early 2026, the open-source project OpenClaw became the most-starred software project on GitHub, surpassing React with over 250,000 stars in 60 days. Created by Peter Steinberger, OpenClaw is a self-hosted, persistent…
Shepherd Model Gateway (SMG) has disaggregated all CPU-bound workloads from GPU inference in large language model serving, moving tokenization, detokenization, and parsing into a dedicated Rust gateway layer that communi…
SQLite remains the right default for local, single-user agent memory, but developers should migrate when sync logic accumulates around the file rather than when it grows too large. Patching SQLite with metadata tables, f…
Ably launched a dedicated AI Transport SDK to solve streaming reliability problems in conversational AI products. The SDK replaces standard HTTP connections with persistent channels that support resumable streaming, mult…
Developer Brian Guthrie released Prospero, a plugin for Claude Code that uses large language models to guide writers through a structured four-phase writing process of interrogation, critique, authoring, and revision. Th…
Anthropic released two new premium Claude Code skills, `/ultraplan` and `/ultrareview`, that generate multi-phase implementation plans and merge-readiness assessments by processing code through a closed, cloud-based pipe…
A 22% throughput improvement and up to 97.5% cache hit rate is achievable on LLM serving clusters by routing requests to GPUs that already hold their token prefixes in KV cache, rather than using round-robin load balanci…
Developers who follow the "leave the campsite better than you found it" principle can now extend that practice to AI coding agents by systematically capturing solutions, updating system files, and verifying learnings. Wh…