Tag
MCP
MCP, or Model Context Protocol, is becoming the common layer that lets AI agents call tools and data sources across maps, email, browsers, and enterprise apps. The key issues are standard connectors, live context, auth, and safe runtime boundaries.
25 articles

AWS MCP Server goes GA with IAM context keys
AWS made its MCP Server generally available, adding IAM context keys, unauthenticated docs access, and sandboxed script execution.

MCP flaw may expose 150 million downloads
Ox Security says an MCP design flaw could expose 150 million downloads and up to 200,000 vulnerable instances.

Claude Code CLI Guide Goes Deep on Hooks and MCP
Blake Crosley’s guide maps Claude Code’s CLI, hooks, MCP, and subagents into a practical system for real software work.

Claude Code Advanced Patterns: 5 CLI Agent Design Tips
Most people use Claude Code for a week, come out with "it's basically ChatGPT with a terminal," and miss its real differentiator: persistent memory and a system that grows with use. This article compresses six months of production use into five patterns that matter — what CLAUDE.md is really for, what to store (and not store) in MEMORY, the composition logic behind Skills, three ways people misuse Subagent dispatch, and the trade-offs between tmux, MCP servers, and file buses for cross-CLI orchestration.

agents-radar tracks AI signals from 10 sources
This GitHub Actions project pulls AI signals from 10 sources daily, publishes bilingual digests, and exposes them through MCP.

MemPalace’s 100% memory claim gets checked
MemPalace hit 11K GitHub stars fast, but its 100% LongMemEval claim fell to 84.2% under compression. The project is real; the marketing isn’t.

n8n 2.14 adds MCP workflow creation
n8n 2.14 adds official MCP workflow creation, and one Reddit user got Claude to build a 13-node workflow in about 2 minutes.

How AI and Maps Are Changing Navigation
MCP is turning maps into AI tools. Baidu, AutoNavi, and Tencent are exposing richer routing, POI, and live data through one protocol.

Kraken CLI brings crypto trading to AI agents
Kraken open-sources a Rust CLI with 134 commands, MCP support, and paper trading so AI agents can trade crypto safely.

Microsoft Agent Framework Adds MCP Tool Options
Microsoft’s Agent Framework now supports MCP over stdio, HTTP, and WebSocket, with runtime headers and tighter credential handling.

OpenAI’s April 2026 updates: Codex gets tighter
OpenAI’s latest Codex release adds proxy-only networking, device-code login, dynamic auth tokens, and more reliable sandbox behavior.

Kiro and AWS HealthOmics Cut Workflow Friction
AWS says Kiro plus HealthOmics can more than double workflow creation speed, with one RNA-seq migration dropping from days to under half a day.

Opera Neon adds MCP support for AI clients
Opera Neon can now act as an MCP server, letting Claude Code, Lovable, and n8n read and control live browser sessions.

PhpStorm 2026.1 Adds Laravel 13 and Agent Tools
PhpStorm 2026.1 adds Laravel 13 support, MCP access for agents, PHP 8.5 pipe fixes, and first-class git worktrees.

CrewForm Agents Can Now Act Like MCP Tools
CrewForm now exposes published agents as MCP tools, so Claude Desktop and Cursor can call them directly with one config file.

OpenCode adds MCP servers with OAuth support
OpenCode now wires local and remote MCP servers into its LLM, with OAuth, per-agent controls, and token-aware guardrails.

What AI Agents Are and How They Work
AI agents combine LLMs, memory, tools, and planning. IBM says they can call APIs, search data, and coordinate tasks autonomously.

Visual Studio adds custom Copilot agents
Visual Studio’s March update adds custom Copilot agents, symbol-aware code search, live profiling tips, and NuGet vulnerability fixes.

Mimosa builds evolving multi-agent science workflows
Mimosa auto-builds and refines scientific agent workflows, aiming to beat rigid pipelines with adaptive tool use and logged execution traces.

Claude Code 里的 Harness Engineering 思路
Claude Code 把 MCP、Skills、Hooks 和 Subagents 直接暴露给用户,展示了 Anthropic 如何把 Harness Engineering 做进产品。

Agent Infra Is Rewriting AI Infrastructure
SWE-agent, Anthropic, and MCP show why agent performance now depends on interfaces, state, and scheduling, not model size alone.

Spec-Driven AI Turns MCP Into a Workflow Engine
MCP, Kiro, and Reaper show how spec-first AI can plan tasks, then execute them with fewer mistakes across creative and business work.

MCP Explained: From Prompts to Production
MCP lets AI apps request data, tools, and confirmations with tighter control. Zoho breaks down how it moves assistants into production workflows.

Google’s 2026 AI Agent Report, Decoded
Google’s AI Agent Trends 2026 report surveyed 3,466 leaders. Here’s what the numbers say about ROI, workflows, security, and hiring.

MCP Standard in 2026: Integrating AI Tools
MCP in 2026: how teams adopt Model Context Protocol for AI tool integration, ensure safe rollouts, and connect hosts with servers.