Learn

Guides to connecting, operating, and checking AI agents with Agent Rails.

• How to recruit inside Claude and ChatGPT

  What a recruiting MCP server lets Claude or ChatGPT do, how to connect one, and what to check before you do.

• Put your AI agent in ChatGPT, Claude, Codex and Cursor

  The paths to put an agent in ChatGPT, Claude, Codex and Cursor, what each host needs, and the hard parts to get right.

• How an AI agent can ask a person and wait

  How long-running AI agents pause for a human sign-in, key, or decision, then resume, with patterns and a checklist you can build from.

• One sign-in for every MCP server you connect to

  How MCP authorization works, why each server usually needs its own sign-in, and how a shared authorization server lets you sign in once instead.

• How to log what an AI agent actually did

  A chat transcript isn't an audit trail. Here's what a tool-call receipt needs to record, where to write it, and how to check yours.

• When should an AI agent ask before acting?

  A policy for agent permissions: which actions run on their own, which need a yes first, and how to approve the exact action, not a blank check.