531 posts tagged with "AI agents"
AI agents and autonomous systems
MCP's '10,000 Connections at Sub-50ms' Claim Is True — and Doesn't Mean What You Think
The 2026 benchmark headline about MCP servers handling 10,000 connections at sub-50ms is technically true and mostly beside the point. Here's what a production deploy-from-chat MCP server actually needs when a fleet of agents, not one chat session, is calling it.
MCP's Tasks Extension Is Built for Work That Outlives a Single HTTP Request — Exactly the Shape of a Deploy
MCP's July 2026 spec ships a Tasks extension so a deploy tool call can outlive a single HTTP request — here's how tasks/get polling beats the hand-rolled deploy_status workaround every deploy-from-chat integration had to invent before it.
Wassette Runs MCP Tools as WebAssembly, Not Containers: What That Actually Buys a Self-Hosted Fleet
Microsoft's Wassette runs MCP tools as WebAssembly Components with deny-by-default WASI permissions instead of a container or a Firecracker microVM. Here's what the microsecond instantiation and policy.yaml grants actually buy a self-hosted fleet — and the GPU access, fork(), and resource-quota gaps that still mean a heavier sandbox for everything else.
Your MCP Server Doesn't Know Who's Deploying. WorkOS's OAuth 2.1 Stack Fixes That
Most MCP servers still trust a bearer token in an env var. WorkOS's May 2026 OAuth 2.1 stack — CIMD, Resource Indicators, and on-behalf-of token exchange — finally answers the question a deploy-authority MCP server can't skip: which agent, acting for whom, just asked to ship to production?
Canine's MCP Server Can Deploy — Coolify's Still Can't: A Tool-by-Tool Audit
Canine's built-in MCP server ships one write tool — DeployProject — where Coolify's stays fully read-only. Here's the exact tool list across Canine, Coolify, and Dokploy, and what Canine's Kubernetes-native architecture actually buys over each.
Devin Outposts: Cognition Shipped a Kubernetes Operator Instead of a Bigger Cloud
Cognition's Devin Outposts keeps the reasoning in the cloud but moves command execution onto your own machines — and shipped a Kubernetes operator to run it. Here's exactly what devin-outpost-k8s deploys and what a Cluster-API fleet needs to add to host it natively.
CVE-2026-24834: The Kata Containers Bug That Turned a Hardware Isolation Boundary Into a Config Default
A container process with nothing but CAP_MKNOD got root inside a Kata Containers microVM meant to be hardware-isolated. Here's the exploit, the exact audit to run against your fleet, and why the fix was a config default, not a KVM flaw.
Pulumi Neo Is GA and Reads Your Terraform State Directly — Here's Why bex's Deploy API Won't
Pulumi Neo reads live Terraform and CloudFormation state directly and executes changes inside it. bex's MCP server calls a fixed, enumerable set of verbs instead. Here's the concrete case for why an agent that can create and destroy production infrastructure needs to know which shape it's actually operating under.
Your AI App Builder Ships Where It Wants: The Open Deploy-Target Gap Behind v0, Bolt, Lovable, and Replit
v0, Bolt, Lovable, and Replit Agent all generate a working app and default to deploying it on their own cloud. Here's exactly what 'export the code' does and doesn't get you, and what it actually takes to run the result somewhere you own.