Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

OAuth for AI agents

OAuth for AI agents is the use of the OAuth authorization standard to let an agent or MCP client act on a user's behalf with a limited, revocable access token, instead of holding the user's password or a broad API key.

Also called: OAuth, MCP authorization, OAuth 2.1

By Divya Sharma · Checked Nov 15, 2026 · How we review

OAuth is the protocol behind "Sign in with" and "Allow this app to access your calendar" screens. RFC 6749 describes it as a way for a third-party application to obtain limited access to an HTTP service, either for a user who approves it or on its own behalf. Agents need exactly this: access to someone's GitHub, Google Drive or CRM, limited to what the task requires and easy to revoke.

MCP authorization

The Model Context Protocol specification defines how remote MCP servers use OAuth. A protected MCP server acts as an OAuth 2.1 resource server that accepts access tokens, and the MCP client acts as the OAuth client. The client discovers which authorization server to use from the MCP server's protected resource metadata; the user signs in and approves scopes there; the client receives a token and sends it with each request. Dynamic client registration lets a new client register without manual setup.

Why it beats a pasted key

A long-lived API key often grants full account access and is easy to leak into logs or prompts. OAuth tokens can be scoped (read issues but not delete repos), expire, and be revoked per client. That supports narrow agent permissions. Hosted MCP servers such as GitHub's offer an OAuth sign-in.

Pitfalls

The MCP security guidance warns against token passthrough, where an MCP server accepts a token not issued for it and forwards it to a downstream API, and describes confused deputy attacks on MCP proxy servers. Scopes requested at sign-in are often broader than a task needs, so review them. OAuth controls access; it does not stop a manipulated agent from misusing the access it has been given. For anything beyond this overview, follow the specification and your identity provider's documentation.

Related terms

MCP server
An MCP server is a program that exposes tools, data resources or prompt templates to AI applications through the Model Context Protocol, so any MCP-compatible client can use them without custom integration code.
API key
An API key is a secret string an application sends with each request so that a service can tell which account is calling, apply its limits and bill it. Anyone holding the key can usually use it.
Agent permissions
Agent permissions are the rules that decide which tools, files, commands, networks and accounts an AI agent may use, and which actions need a human's approval first, so a mistaken or manipulated agent can do limited damage.
Model Context Protocol
The Model Context Protocol (MCP) is an open standard that defines how AI applications connect to external tools, data and prompts, using JSON-RPC messages between a host application, its clients and separate servers.

Sources