Thank you so much to our Ruby Friends of BSidesSF!
@Coffeetocode - Enso-Remco - Enso-Edward - David Shipley - Adam Arellano - bykhee - frostbide - Dave Herrald - Chris Palmer
WOW! Y'all are truly fantastic!
bsidessf.org/friends#bsidessf#BSidesSF2025#infosec
π¨βοΈ EMERGENCY COFFEE PLAN ACTIVATED! βοΈπ¨
Our beloved coffee machine decided to take a break π Going old school with a hot water pour-over method! π₯π«
Stay calm, caffeine is on the way. πͺβ¨
#CoffeeCrisis#StayCalm#CaffeineFix#CoffeeToCode
They didnβt exit though. They are hedging bets. Most folks leave Mastadon when the network effect isnβt there. We have been through this cycle before.
On secret/sensitive wrapper types:
Iβve seen how difficult it is keep secrets or sensitive data from entering various logging systems. When it gets in there your security team is now busy cleaning log data instead of what is most helpful to the org.
On uuids: this doesnβt prevent IDORs but does help make them more difficult to exploit. Additionally helps with reducing information leaks via competitive intelligence (βoh im user 173648β instead of <uuid here>)
some case specific examples: elimination of cryptographic agility (see: wireguard), logic-less templates
@travismcpeak's thread that @leifdreizler mentioned:
Webauthn, memory safe languages, only role assumption on aws (no iam users), functions that only accept compile time constants as format strings, ORMs, csp, binary safelisting; off the top of my head π
We have a react link component that has an allowlist of schemes, mostly to stop js hrefs.
We also have a test that uses "Jest spies" to try and catch when someone forgets an authZ when accessing a DB table
Yep! I mean the best defense is to ensure you just donβt have XSS in the first place, but service workers can be a nice defense in depth approach to protect sensitive data