AI Agents news and analysis on Web Pulse: 49187 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
A 24 September 2026 arXiv paper reports that Nubank used the Snowglobe simulator to screen candidate customer-experience agents before deployment, cutting AI agent update cycles from 3 weeks to under 5 days. Simulation-g…
Taskivities, an MIT-licensed open-source task manager, launched on Hacker News' Show HN with a design rule that every task must carry a project, a date, an owner and a priority, and overdue work rolls into today marked w…
A developer behind StateLode, a hosted task board that coding agents read and write over MCP, documented five ways real MCP clients broke the server after its Streamable HTTP endpoint went live, including a GET request f…
GitHub added a `copilot_code_review` object to every `repos-1-day` row in its enterprise and organization Copilot usage metrics reports, letting teams compute an addressed rate per repository by dividing `total_comments_…
Taskhold measured the context-window cost of its own MCP server at roughly 8,000 tokens across 22 tools, and cut it to about 7,200 by replacing the 180-character regex that zod's .uuid() writes into all 20 id fields with…
Prisment launched as a free, local-first document workspace that positions itself as a verification layer between human intent and AI-generated software, letting users open an entire project folder and review AI-drafted …
An analysis of 2,688 public Claude Code plugins across 1,961 active repositories found that nearly a third run code automatically when a session starts, and that most MCP servers launched via npx or bunx pin no version. …
A developer writing on the blog hoelee.com documented how a self-hosted OpenAI-compatible model gateway caused six independent workflow seats to be disabled simultaneously after a vendor trust-and-safety block, because t…
Tencent released Hy4, a 770-billion-parameter open-weight model with visual understanding and computer-use capabilities, on November 4, 2026, claiming a 9-of-12 scenario lead on its WorkBuddy evaluation and a finance sco…
A guide drafted by an AI agent outlines how estate agents and letting agencies should deploy property chatbots under the EU AI Act's Article 50, requiring an AI disclosure at the first message and grounding answers about…
A developer built an agentic loop on Claude Code that iteratively rewrites AI-generated text to evade Pangram's AI-text detector, using the detector's score as an objective function to hillclimb toward 0% AI. After five …
Thinking Machines introduced Inkling 2 Small, a Mixture-of-Experts model with 594B total parameters and 17B active, scoring 31 on the Artificial Analysis Intelligence Index and initially available through Tinker with clo…
AMD's ROCm InferaSim replaces most LLM serving-configuration measurements with fast analytical projection, measuring once per set of kernels and letting an agent search projections that each take under half a second on o…
A 6 November 2026 deployment audit and pricing review found that the same MCP website-search service costs about US$7 per month at 100,000 searches in a minimal single-region Azure Container Apps design, versus roughly U…
The open-source Teardown skill for Claude Code, created October 30, 2026, builds a single offline HTML 3D simulator of a machine from a plain-language request, and its GitHub repository (elvarofficial/teardown) had about…
Cursor Team Rules, available on Cursor's Team and Enterprise plans, let a team admin write rules once in the Cursor dashboard that Cursor then adds to Agent for every member in every repository, with an admin able to mar…
Check Point Research reported on November 5, 2026 that it identified 2,557 systems that hosted ARTEX panels, the open-source AI penetration-testing platform, with 1,608 live and publicly reachable at research time, and r…
The open-source GitHub repository rea-video, an agent skill built to run inside Claude Code, Codex or any command-capable agent, has drawn about 2.2k stars and 318 forks under the Apache-2.0 license, with a trending coun…
Anthropic's privacy hook in Claude Code v2.1.317, enabled by default, fails to block 8 of 23 email-leak shapes tested by Frank Chu, with a reproduction on v2.1.318 filed as GitHub issue 107757. The hook, cc-plugin-privac…
The ArchDocs team published its methodology for a November 5–6, 2026 study testing whether coding agents using CLAUDE.md or AGENTS.md rules files miss lessons already recorded by the team, and whether ArchDocs' get_task_…