# Start from an API key. End able to read any flow. 6 short lessons, about 40 minutes in all. They assume you have built things that call APIs, and nothing else: no OAuth, no JWTs, no specifications. ## The lessons In order. Each builds on the one before, and each ends with questions to check yourself. 1. [**What does the API actually see?** Say what an API can and cannot know about a request from an agent, and why an API key is not enough. 5 min · Credential, API key, Client, Resource server](https://whosonfirst.ai/learn/what-the-api-sees/) 2. [**What is a token?** Read an access token, name the five checks an API runs before believing it, and say what a valid token still does not prove. 7 min · Access token, Authorization server, Claim, Issuer, and more](https://whosonfirst.ai/learn/tokens/) 3. [**How does an agent get a token for a user?** Explain each hop of an OAuth sign-in: why the browser is involved, why a code comes back before a token, and what a refresh token is for. 8 min · OAuth, Consent, Authorization code, PKCE, and more](https://whosonfirst.ai/learn/getting-a-token/) 4. [**What if nobody is there to sign in?** Decide whether an agent should act for a person or as itself, and say where an agent’s own identity comes from. 7 min · Principal, Service account, Client credentials, Workload, and more](https://whosonfirst.ai/learn/whose-authority/) 5. [**What happens when the agent calls a tool that calls an API?** Explain why a token should be exchanged at each hop and not passed along, and tell delegation from impersonation. 6 min · Token passthrough, Confused deputy, Token exchange, Delegation, and more](https://whosonfirst.ai/learn/chains/) 6. [**What limits the damage when something goes wrong?** Name the two failures every agent design has to plan for, and the controls that limit each. 7 min · Prompt injection, Sender-constrained token, Proof of possession, DPoP, and more](https://whosonfirst.ai/learn/limits/) [Start with lesson 1 →](https://whosonfirst.ai/learn/what-the-api-sees/) ## How the lessons work - **One example all the way through.** Alice, an agent, and a calendar API. Each lesson changes one thing about it. - **Plain words first.** The specifications’ terms arrive after the idea they name, and each lesson ends with a table of them. - **A figure you can change.** Most lessons have one. It shows a complete answer before you touch it, and a table beneath it covers every case, so nothing depends on clicking. - **Nothing to install.** These lessons explain the model. They are not a tutorial for one product, and every page they point to says what it was checked against. ## Where to go from here After the lessons, or instead of them if you know this ground already. ### [Choose a flow for your agent](https://whosonfirst.ai/guide/) Six questions about your agent, in the words these lessons used. The answer is a short list of flows, with the reason for each. ### [Read a flow at full size](https://whosonfirst.ai/flows/) 20 flows as step-by-step diagrams, with the HTTP on the wire. Start with authorization code with PKCE: it is lesson 3 with nothing left out. ### [See what changes where your agent runs](https://whosonfirst.ai/deployments/) Hosted apps, Kubernetes, managed agent platforms, desktops and command lines, CI and sandboxes: where the first credential comes from in each. ### [Keep the whole model on one page](https://whosonfirst.ai/concepts/) The same ideas as these lessons, compressed, with the caveats left in. Useful once you know the ground. --- From Who’s on First, a field guide to agent authentication and authorization. This page: https://whosonfirst.ai/learn/ Standards status last reviewed November 2026. Drafts move quickly: check the linked source before you build on one.