§Filed under · Tag archive
MCP.
03 posts
- 01
How AI Agents Talk to External Systems: Six Patterns and When to Use Each
The model never touches the network. It emits a request and your harness acts. Six integration patterns answer what sits on the other side, and each one sets who owns the interface, where data lives, and how far a mistake can reach.
AI AgentsFunction CallingAgent ArchitectureA2A33 mins read - 02
MCP Server Authorization: The Token You Must Never Forward
A valid token is not permission. Here's how OAuth works for MCP servers, why scopes don't replace object-level checks, and the passthrough mistake that turns your server into a confused deputy.
SecurityOAuthAI AgentsAPI Design22 mins read - 03
MCP Server Design Is Changing: From API Endpoints to Agent Workflows
One-to-one API-to-tool mapping is the easiest way to build an MCP server and one of the easiest ways to build a bad one. Here's the two-layer pattern replacing it.
AI AgentsAPI DesignAnthropicAgent Architecture17 mins read