Threat Modelling Stories From The Trenches
summary
Six concrete failure stories from real production systems reveal where controls fail in practice: an mutual TLS vs XML signature mismatch that lets a client claim any sender in logs, a step‑up session token substitution that lets attackers “step over” MFA, a hashing fix that regresses to cleartext passwords in persisted sessions, OTP reuse across unsynced instances due to load balancers, SIM/APN devices giving direct IP access through an open firewall to the backend, and user ID namespaces abused to lock rivals out of leaderboards.
AI-written summary · about
chapters (22)
- 0:00Intro
- 1:29What is Threat Modelling?
- 2:53What is a Design Flaw?
- 4:40Who Sent the Message?
- 4:56In Theory: Good Security Controls
- 6:17In Practice: Design Flaws in Integration
- 8:51Mind the Gap: Stepping Over the Step-Up
- 9:06Step-up Authentication
- 11:16Attacking Step-up
- 18:46Secure Password Storage with Insecure Side Effects
- 20:32Developers Come to the Rescue
- 21:09Production Web Farm Using Session DB
- 23:26One-Time Passwords in Parallel Universes
- 23:36Multi-factor Authentication for Remote Access
- 25:11Side-effect of Load Balancing Servers
- 27:29Secure Connection or Open Door?
- 29:32Attacking the System
- 32:37Attack the Datacentre
- 34:42What is an Asset?
- 36:01Gaming: Is Userld an Asset?
- 36:32How Can Bob Climb the Leaderboard?
- 38:56Summary
related talks
45mMFA: Stories that make you go 'Huh?'
29mThreat Modelling 101
45mA door with no locks? Let’s talk about threat modeling
28mLearning from High Profile Failures
30mTales of Red Teaming
51mSecure Applications, By Design
14mThreat Modeling for O11y: From Blind Spots to Actionable Insight
47mHacking Applications on Paper