Web security with the browser’s entire safety net removed and the data served raw to untrusted machine clients. APIs are how modern software actually talks — phone apps, SPAs, microservices, third-party integrations — and OWASP maintains a separate API Security Top 10 because the threat profile is genuinely different from web apps.
The one idea: an API must enforce, on every stateless request, everything a browser used to enforce for free — and its defining weakness isn’t injection but authorization: the API knows who you are and forgets to check what you’re allowed to touch. Three of OWASP’s top five API risks are that one failure.
Why this exists — and what it doesn’t re-teach
The vault covered web security and backend auth, but API-specific security was a gap — despite APIs being where most modern attacks land. This folder covers what’s specific to APIs and cross-links the rest: it doesn’t re-teach injection or OAuth from scratch, it shows how they change when the client is a machine and the browser is gone. It’s the attacking-and-defending counterpart to API design.
Reading order
Built around the OWASP API Security Top 10 (2023). Read 01 for the map, then 03 is the heart of it.
- what-makes-api-security-different — [Intermediate] — the protections APIs lose, why machine clients change the threat model, and why authorization dominates. The OWASP API Top 10 as a map
- authentication-for-apis — [Intermediate → Advanced] — API keys, JWT (and its recurring pitfalls), OAuth/OIDC, mTLS, and the token-storage problem
- authorization-and-bola — [Intermediate → Advanced] — the #1 API vulnerability (BOLA/IDOR), plus BFLA, mass assignment and excessive data exposure: one mistake wearing four hats
- input-validation-and-injection — [Intermediate → Advanced] — schema validation, the injection classes on an API surface, and SSRF — the API era’s signature vulnerability (Capital One)
- rate-limiting-and-abuse — [Intermediate] — machine-scale abuse, resource consumption, and business-flow abuse that uses the API exactly as designed (which a rate limit alone can’t catch)
- the-api-security-lifecycle — [Intermediate] — shadow/zombie APIs, misconfiguration, consuming third-party APIs safely, gateways, and testing (why BOLA needs manual testing)
If you only take three things
- Authorization is the #1 problem — scope every query to the user and check ownership on every object, because no framework does it for you (03).
- The client is untrusted — a mobile app or SPA runs on the attacker’s device, so every check must be re-done server-side (01).
- When a user supplies a URL, assume SSRF — it can reach your cloud metadata and internal services (04).
Practice
- PortSwigger Web Security Academy — has dedicated API testing, access control (IDOR/BOLA), and SSRF labs. Free, and the best place to build this → cybersecurity projects
- crAPI (OWASP’s deliberately-vulnerable API) and VAmPI — purpose-built API-security practice targets
- Burp Suite (+ the Autorize extension) and Postman — the workhorse tools for testing your own APIs → tools
Related
- cybersecurity curriculum map · web security — the parent discipline
- API design · backend/auth — the building side
- ethical hacking — API testing as modern web pentesting
- cloud security — the SSRF-to-metadata path
Software engineering
Infrastructure & security
Computer science
Hardware & engineering