266 posts tagged with "Model Context Protocol"
Content about the Model Context Protocol (MCP)
MCP Goes Stateless: What Losing Long-Lived Connections Costs — and Buys — the Servers You Self-Host
MCP's 2026-07-28 revision deletes sessions, the initialize handshake, and every server-to-client primitive. A gains-versus-costs accounting for operators of self-hosted MCP tool servers, plus the task-based redesign for deploys that outlive one request.
Your MCP Server Didn't Break — Its SDK Moved: Surviving MCP Python SDK 2.0's FastMCP-to-MCPServer Rename
MCP Python SDK 2.0 renamed FastMCP to MCPServer with no shim, so every fresh install against an unbounded mcp>=1.x pin fails at import. We cover the five-minute pin, the full port including four gotchas past the rename, and the fresh-resolve CI job that catches the next breaking major.
MCP Went Stateless: What Killing the Session Means for an Endpoint That Deploys Apps From Chat
MCP's 2026-07-28 spec removes protocol-level sessions and the Mcp-Session-Id header entirely. The before/after deployment table — sticky ALB plus session store versus plain round-robin — plus the five things stateless does not fix and what a deploy-from-chat endpoint must change to survive agent load.
Cloudflare's Agents Week 2026 Is a Requirements Doc for Agent Infrastructure: A Self-Hosted Reading
Cloudflare's August 2026 Agents Week maps, day by day, everything AI agents need from infrastructure — per-task computers, machine-readable APIs, per-agent identity, and a way to pay. Each item translated into what it takes to run it on machines you own, with an honest ledger of what shipped versus what is still beta.
One Token, Seven Tools, Total Access: What a Czech Community's Coolify MCP Bridge Teaches About Agent-Operated Self-Hosting
A Czech community's 200-line MCP bridge lets AI agents operate Coolify apps from AnythingLLM behind a single bearer token, no permission model, and TLS verification off. A code-level inventory of its seven tools and the guardrail checklist that separates homelab experiments from tenant-ready agent platforms.
A Chip Vendor Put Its Catalog Inside Your AI Assistant: What Microchip's MCP Server Teaches Every Platform API
Microchip's MCP server lets any AI assistant answer engineering questions from verified parts data. The five things it does that every platform API must copy — and the read-only-vs-read-write caveat that decides whether your agent interface is safe.
Netlify Calls the Model for You: What Zero-Key Agent Runners Teach Self-Hosted MCP Servers About Auth
Netlify's Agent Runners remove model API keys by billing inference through its AI Gateway; why a self-hosted deploy-from-chat platform should answer with tenant-held keys plus short-lived OAuth 2.1 agent credentials instead of rebuilding the broker.
Harvest Now, Decrypt Later Comes for Your MCP Server: Why Agent Deploy Calls Are Worth Storing
A June 2026 executive order made harvest-now-decrypt-later federal policy — and the ciphertext most worth storing is your agent's deploy calls. A three-layer exposure audit plus a concrete checklist for hybrid post-quantum key exchange on an infra MCP server's own transport.
No Human Typed This Deploy: What Claude's Cron-Triggered Agents Demand From Your MCP Auth Story
Anthropic's June 2026 scheduled deployments let agents wake on a cron schedule, pull vault credentials, and ship with no human watching. Here is the six-rule auth checklist — scoped per-run grants, platform-written audit trails, blast-radius limits — a deploy-from-chat MCP surface needs before a timer gets deploy rights.