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.
OpenAI announced on September 8th, 2026 that a fleet of 10,000 AI agents running GPT-6 Astra and an unreleased model, at a cost of millions of dollars, discovered an analytical solution to a specific question about the N…
Glow Labs' Glow Security research team published a study, MCPhantom, finding that coding agents asked to set up an MCP server installed deprecated, archived servers in 98% of tracked sessions, pulling package names from …
A developer described using the open-source Superpowers toolset to spec out features with a coding agent before implementation, working through a runbook-in-design-doc example to surface questions about version history, …
JetBrains open-sourced ThinkRail, an agentic IDE that runs each AI agent session in its own git worktree, under Apache 2.0 for macOS, Windows, and Linux. ThinkRail is built on Earendil Works' MIT-licensed Pi coding agent…
Tencent Holdings released its flagship large language model Hy4 on Wednesday, adding visual understanding for the first time so the model can process images, analyze complex charts and perform visual mathematical reasoni…
Keystone shipped a Claude Code plugin that installs with one command and provides eight skills plus two MCP servers for deploying and integrating its LLM gateway, with the /ktms:assess skill reading a repository to map e…
A measured Claude Opus 5.5 agent session of 35 calls cost $2.39 at Anthropic list prices because 97.5% of prompt tokens were cache hits, versus $22.08 without prompt caching — a 9.2x difference for identical tokens, acco…
A developer released readfirst, an MIT-licensed tool that stores team knowledge about a codebase in a single `.readfirst/map.json` file and serves it file by file to coding agents such as Claude Code, Cursor, Codex and G…
Malik Ali Hassan built BijliCheck, a framework-free web app hosted on Vercel that computes a Pakistani electricity bill first and then lets a RAG-based assistant answer plain-English questions using only the user's compu…
A developer turned DeepSeek Harness's design essay on super-agents into a build guide, outlining a two-layer architecture with a long-lived Router session that handles human interaction and dispatches work to isolated Wo…
Alibaba Cloud released CMS MCP, a Model Context Protocol server that maps the leaf commands of its `aliyun cms2` CLI one-to-one onto MCP tools, letting AI agents query alerts, run PromQL range queries, pull trace trees a…
A developer published a design guide for production-grade Model Context Protocol (MCP) toolchains that insert deterministic infrastructure — a policy engine, tool gateway, and sandboxed worker — between an agent's intent…
A new analysis finds that long-horizon terminal-task benchmarks commonly measure task length by the number of turns a model needs to pass and difficulty by labels an LLM assigns during generation, and the authors report …
StartGlobal-AI released AgentFirstStack, an MIT-licensed, open-source bundle of analytics, tag management, email inboxes and session replay that an AI coding agent installs by making HTTP requests, with no signup form, v…
AyeWatch, a page-watching service from ayewatch.ai, monitors a hotel booking page for specified dates and sends an alert when the rate crosses a user-defined threshold, with paid plans starting at $9 a month as of Novemb…
A Google Developer Expert for Firebase published a hands-on breakdown of Firebase AI Logic, which lets web and mobile clients call Gemini and Imagen models through the Firebase Web SDK while keeping API credentials off t…
Shell Drop released a free iOS terminal app that lets users attach to tmux sessions on a remote Mac mini or VPS over SSH and steer Claude Code and other terminal-based agents from an iPhone. The app saves connections for…
The Firebase team released streaming responses for Cloud Functions callable functions and a new onCallGenkit trigger, letting server functions push data to clients in chunks as it becomes available and integrate Firebase…
An October 8 analysis by Engram tested 6 memory systems on 1,536 LoCoMo questions, evaluating retrieval and answer accuracy separately, and found that retrieval success does not guarantee answer correctness. The Agent Me…
OpenCode 2.2.5, the MIT-licensed AI coding agent from Anomaly, shipped on October 30, 2026, and a new guide documents installing it via curl script, npm (@opencode/cli) or Homebrew, connecting models from more than 75 pr…