AI Agents

MCP server security guide: use Snyk’s MCP-Scan to catch poisoned tools and Docker MCP Gateway to sandbox and allowlist what your AI coding agents can run.

Kite AI’s mainnet puts x402 stablecoin rails into production for AI agent payments. We tested spend caps, agent identity and fraud controls merchants still lack

CrowdStrike Charlotte AI now runs SafeMind models and 50+ Falcon IQ agents. See how agentic SOC triage works in 2026 and what it means for security teams.

Wire Windsurf Cascade to Microsoft’s Playwright MCP server: exact config file, working prompts, Browser Preview auto-QA loop, and the failure modes nobody warns you about

Most agent breaches in 2026 don’t target the model — they hit the plumbing. Here’s where security actually belongs in your AI agent security stack, layer by layer

Motif Deep Research is a $12/mo agentic search agent running multi-hour, citation-locked research loops — the same job ChatGPT Pro and Perplexity Max charge $200 for

Straiker’s $64M Series A signals AI agent security is now its own funded category — what changed in 2026 and how to budget for prompt injection defense

The MCP 2026-07-28 spec kills Mcp-Session-Id, drops initialize, deprecates three features and rebuilds auth — what remote server operators must migrate this week

Chrome DevTools MCP gives AI coding agents live access to the console, network, DOM, and performance panel — here’s how to install and use it in 2026

Cloudflare Code Mode MCP turns MCP tools into a generated TypeScript API agents write code against — cutting 60-90% of tokens on multi-step chains

Mem0, Letta and Zep all shipped major releases this month with clashing benchmark claims. We tested all three to find which AI agent memory layer actually holds up in production

AI shopping agents now buy without ever loading your product page. Here’s how the agentic commerce protocol, ACP and AP2 work — and how to get your catalog picked in 2026.

AI’s MCP 2026-07-28 specification ships July 28 with a stateless core that runs behind load balancers, server-rendered Apps in sandboxed iframes, a Tasks extension for long jobs, and OAuth/OIDC auth

Kimi K2 Thinking is Moonshot’s trillion-parameter open-weight model that rivals GPT-5 and Claude on agentic and coding benchmarks for a fraction of the price

MCP for Enterprise 2026: 15-chapter playbook on production MCP servers, OAuth 2.1 auth, gateway federation, audit, transport scalability, vendor landscape.

Multi-agent systems in 2026: 24-chapter playbook on topologies, coordination, memory, tools, evaluation, observability, cost, security, and deployment.

Build production AI agents with LangGraph and MCP: graphs, multi-agent supervision, checkpoints, observability, deployment, cost, and operational patterns.

Verkor.io’s AI agent Design Conductor produced a working RISC-V CPU core (VerCore, 1.48GHz) from a 219-word spec in 12 hours. The agentic chip design milestone of 2026.

Perplexity launched Finance Search in its Agent API on May 6 — one tool call for licensed financial data, real-time prices, and cited sources.