L18 Authentication and authorization
21 minSuggested first: L17 Timeouts and retries
Outcome
Identify the evidence for identity and the evidence for permission at a resource boundary.
Learn
Authentication establishes who is making the request. Authorization decides whether that identity may perform this action on this resource. Hiding a button in the browser is a convenience, not enforcement.
When you read the code, follow the identity from a verified session to the server-side check on the resource.
Worked example
Three people ask for one document. Who gets which answer?
session invalid → 401
session valid and user is not owner → 403
session valid and user is owner → read document
Predict first, then show the worked explanation
A missing or invalid session gets 401; a valid user who doesn't own the document gets 403; a valid owner reaches the read. This assumes the document exists and the authentication protocol supplies whatever challenge a 401 needs. Some production systems deliberately answer 404 to hide that a resource exists.
Practice and check
Common mistake
Treating a user id from the request body as a verified identity. Trace the trusted session, and check that the ownership rule applies to the resource being read.
Practice on a real repository
Try the puzzle below to look at protection at a different boundary, credential forwarding.
- Dropping credentials on cross-origin redirectsFind · Medium · encode-httpx
Related topics: Auth and security
Practice is optional. Solving a puzzle doesn't complete a lesson, and lessons never change a puzzle's points.
References
- RFC 9110: HTTP Semantics (IETF)
- Authorization Cheat Sheet (OWASP)