MCP guides and AI agent security
AI agent authentication: methods and access control
AI agent authentication explained: API keys, OAuth, scoped tokens and per-tool access control, and which method fits personal agents and teams.
Updated 2026-10-14
AI agent identity and non-human identity
AI agent identity explained: how non-human identities work, how to manage them, and how to give each agent its own credentials and permissions.
Updated 2026-10-14
AI agent integration: how to connect agents to your apps
AI agent integration options: native connectors, MCP servers, gateways and custom APIs. How to connect an agent to Gmail, Notion or Jira safely.
Updated 2026-10-14
AI agent observability: best practices and tools
AI agent observability best practices: what to log, tracing tool calls, dashboards and tools, and how an audit log answers which agent did what.
Updated 2026-10-14
AI agent security: risks and governance basics
AI agent security basics: the main risks, governance frameworks and controls that work for small teams, from scoped access to audit logs.
Updated 2026-10-14
AI agent tools: the best tools for AI agents
AI agent tools worth using in 2026: frameworks, MCP servers, gateways, memory and observability, with what each one is for and what it costs.
Updated 2026-10-14
MCP server tutorial: how to build an MCP server
MCP server tutorial: build a working MCP server step by step, test it in Claude and Cursor, then host it remotely or put it behind a gateway.
Updated 2026-10-14
LLM gateway vs MCP gateway
LLM gateway vs MCP gateway: one routes model calls, the other routes tool calls. What each does, Cloudflare AI Gateway as an example, when you need both.
Updated 2026-10-14
MCP architecture: hosts, clients and servers
MCP architecture explained with diagrams: hosts, clients, servers, transports and how a tool call travels from the model to your app and back.
Updated 2026-10-14
MCP client: what it is and how it differs from a server
An MCP client is the part of an AI app that talks to MCP servers. How MCP clients work, MCP client vs server, and a list of popular clients.
Updated 2026-10-14
MCP security best practices
MCP security best practices: prompt injection, token theft, over-broad scopes and tool poisoning, with concrete fixes for Claude, Cursor and Codex setups.
Updated 2026-10-14
MCP specification: a plain-English guide
MCP specification in plain English: versions, transports, auth and security rules of the Model Context Protocol, and what changed in each release.
Updated 2026-10-14
MCP tools, resources and prompts explained
MCP tools, resources and prompts: what each primitive does, MCP resources vs tools, and how clients like Claude and Cursor use them in practice.
Updated 2026-10-14
Tool calling vs MCP: what is the difference
Tool calling vs MCP: function calling is how a model asks for a tool, MCP is how tools are shipped to any agent. Differences, examples, when to use each.
Updated 2026-10-14
What is an MCP server?
What is an MCP server: a program that exposes tools and data to AI agents over the Model Context Protocol. How it works, examples, local vs remote.
Updated 2026-10-14
What is MCP (Model Context Protocol)?
What is MCP: the Model Context Protocol is an open standard that connects AI models to tools and data. How it works, who supports it, and examples.
Updated 2026-10-14