ChatGPT Voice on Desktop: Hands-Free Agent Control
OpenAI brought ChatGPT Voice to the macOS and Windows desktop app, letting you direct Codex and ChatGPT Work agents hands-free with GPT-Live full-duplex voice.
Claude Managed Agents Now Mount 500 Skills a Session
Anthropic raised Claude Managed Agents from 20 to 500 skills per session and added five-tier effort controls, making a single agent able to hold a whole knowledge base.
Microsoft Ships the Agent Framework Harness
Microsoft released the Agent Framework Harness on July 22 2026: a batteries-included agent runtime for Python and .NET that bundles the loop, memory, skills, planning and approvals into one component.
Subagentmaxxing: Cheap AI Coders as Claude Subagents
Subagentmaxxing is a new open-source CLI that runs Codex, Cursor, and Grok as subagents under Claude Code, keeping flagship planning while cutting execution costs up to 25 times.
AgentPrizm Gives Claude Code, Cursor Persistent Memory
AgentPrizm launched AgentMemory and AgentSkills, giving AI coding agents persistent, auditable memory and a marketplace of reusable, governed procedures over REST API and MCP.
Kastra Adds Guardrails to Claude Code, Cursor, Codex
Kastra is a runtime authorization layer that checks every shell command, file edit, and API call your AI coding agent makes before it runs.
Devin Security Swarm: AI Agents That Fix Vulnerabilities
Cognition's Devin Security Swarm scans code with parallel agents, proves each exploit in a sandbox, and opens a fix PR. It caught 72% of 50 real GitHub Security Advisory CVEs.
xAI Voice Agent Builder: No-Code Grok Voice Agents
xAI's Voice Agent Builder turns a plain-English description of a phone call into a live agent on Grok Voice in about two minutes, all-inclusive at $0.05 per minute.
Comfy MCP: Run ComfyUI Image and Video From Any Agent
Comfy Org's new public-beta MCP server lets Claude, Cursor, and other agents drive ComfyUI's full image, video, audio, and 3D pipeline in plain language, with no nodes and no local GPU.
Sakana Fugu: One API to Orchestrate Top AI Models
Sakana Fugu is a multi-agent system that behaves like one model: send a request to a single OpenAI-compatible API and it routes across frontier models for you. Here is what it costs, how it scores, and why orchestration matters for builders.
Locaible Runs Local AI Agents Inside Cursor
Locaible runs pre-tuned AI coding agents on your own machine and exposes them to Cursor through a local, OpenAI-compatible endpoint, so your code never leaves your machine.
Xcode 27 Agent Skills Now Work in Claude and Cursor
Apple's Xcode 27 ships seven official Agent Skills, and one command exports them into the ~/.agents folder that Claude, Codex, and Cursor already read.
Cohere North Mini Code: Open 30B Coding Model
Cohere released North Mini Code 1.0 on June 9, 2026, an open-weights 30B Mixture-of-Experts coding model that runs on modest hardware under Apache 2.0.
Cursor Canvas Adds Design Mode and Context Usage Report
Cursor shipped two canvas features on June 4: Design Mode for annotating UI elements directly, and a Context Usage Report that audits where the agent spends its tokens.
Project Solara: Microsoft's Android OS for AI Agents
Microsoft unveiled Project Solara on June 2, 2026, a new Android-based platform built for AI agent interactions rather than traditional app navigation.
Runway MCP: Generate Images and Video in Any MCP Agent
Runway is now an MCP server. You can call its image and video models from inside Claude Desktop, Cursor, and any other MCP-compatible agent.
AWS Strands Agents + NVIDIA NIM: Production Multi-Agent AI
AWS published a reference architecture on May 26, 2026 combining Strands Agents, NVIDIA NIM, and Amazon Bedrock AgentCore into a production-ready multi-agent system.
holaOS 0.1: AI Workstream Manager for Creators
holaOS 0.1 ships Dashboard, Sub Agents, and Multi Workspaces for managing parallel AI workflows on macOS.
Holotron-12B Ships Open-Source Computer Use at 2x Speed
H Company released Holotron-12B on March 17, an open-source computer use agent model that delivers more than 2x the throughput of its predecessor on a single H100 GPU.