AgentDojo / The manual

Good systems start with good contracts.

Notes on connecting tools, defining permissions, and operating agents with clarity.

Tool design
Tool design · 1 min read

Design a tool profile around a task.

Start with what the agent needs to accomplish, then choose its capabilities.

Read the guide
Connections
Connections · 1 min read

Bring private tools into the loop.

Give internal systems an explicit place in your agent architecture.

Read the guide
Operations
Operations · 1 min read

A useful log tells you what happened.

Make agent execution understandable to the next person investigating it.

Read the guide
MCP engineering
MCP engineering · 9 min read

What Is an MCP Gateway? Architecture, Security and Deployment

An MCP gateway is the policy and routing layer between AI agents and the MCP servers they use. It keeps tool access observable, authenticated, and constrained while teams connect more servers.

Read the guide
MCP engineering
MCP engineering · 10 min read

How to Host a Remote MCP Server in Production

A remote MCP server needs more than a public URL. Production hosting requires deployment discipline, OAuth, secret isolation, scaling, monitoring, and a clean example your agents can actually call.

Read the guide
MCP engineering
MCP engineering · 11 min read

How to Build an MCP Server: Complete Developer Guide

To build an MCP server, start with one useful tool, define the input schema, add authentication, test the call path, and deploy it behind a platform that can govern production usage.

Read the guide