ShrekOverflow (@ShrekOverflow) header
ShrekOverflow (@ShrekOverflow) profile picture

Usually okay to talk about security things. Product @workos, prev. architect @Auth0. sometimes I technobabble sarcastically

Bellevue, WA
Joined January 2014
Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)Photo by ShrekOverflow (@ShrekOverflow)
I wanna talk to people want to talk about how they are rolling out agents in their orgs. The goods, the bads, the uglies!
1
1
8
353
This attention to detail is delightful. Thanks @OpenAI! So much more human
111
Trying to fix a throughput bug on node containers, and realized I'm running into Fargate bandwidth limitations.
What's the spookiest bug you've ever seen?
103
ShrekOverflow retweeted
The next trillion dollar company will be in cybersecurity. Not in years. In months.
120
151
43
2,401
92,813
Holy shit
Doing some small play mini games Deepseek vs Opus 5.5. This is a play on Starfox as a small play. It's absolutely crazy how far Open models have come.
1
2
238
A lot of companies are going to realize cyber security is actually cyber defense and that makes them
6
268
It’ll soon be time to allowlisting JavaScript by site
65
Perfect security requires omniscience
1
194
The auth people are touching grass! Is this a new OAuth 2 spec that I missed out on?
2
4
302
Note that not everyone updates. So even if you ship an OS patch, an early adopter decompiling and reverse engineering can abuse the leverage. Now that said, this same attack also exists with models far weaker than GLM-5.3, like Opus 4.1 levels should be sufficient.
GLM-5.3 can exploit the period between a security fix becoming publicly available and the fix reaching end users (the patch gap) by analyzing the security fixes it can reverse engineer the underlying vulnerabilities and develop working exploits before users receive the patches
1
169
Need 15-30 minutes max, keen to hear about how ya'll are thinking about what the agent (s) gets accesss to, how, is it audited, what are the teams wanting?
26
Claude just dumped my entire (dummy) 1p vault. So yeah very secure
2
5
226
If San Francisco is X, then Seattle is LinkedIn.
2
9
357
ShrekOverflow retweeted
Honored that @WorkOS has been named to this year’s Cyber60 list 🚀 lsvp.com/cyber60-2026-2027/ Thank you @FortuneMagazine @AWS @lightspeedvp
3
3
30
1,232
Been on this bandwagon since 2023. An ai agent with access to same tools as an employee, is more dangerous than an employee.
In a single year, AI has rewritten the security priority list for CISOs. Attackers are moving faster. Bain estimates AI compresses a typical cyberattack from about four weeks to about 18 hours, and models keep getting better at finding and exploiting weaknesses. There's also more to protect, with employees shipping apps they couldn't have built a year ago and agents getting access to source code, documents, and production systems. In a survey of 100 CISOs at companies with 1,000+ employees, all seven of the most-selected 2027 priorities involve AI. The biggest jumps since 2026: - Internal AI agent security (+28 pts) - AI SOC (+26 pts) - AI offensive security (+24 pts) - AI code security (+23 pts) - AI vulnerability remediation (+22 pts) Last year's staples: - Exposure & vulnerability management (-21 pts) - SIEM / security data (-18 pts) - Privileged access management (-18 pts) From a16z's @OwenColby and @steph_zhang on The New Security Agenda: a16z.news/p/the-new-security…
1
2
140
Okay apple is gone crazy so there is MacBook Pro and MacBook Pro-er? Both can have M5 pro and max? But only one comes with the glow on the back 😂
1
100
I am giddy looking at my screen right now 😬. I feel like I am on a mission
61
Dear lord I just submitted a brief for a talk in 2026, not yapping all about selling shovels to ai agents!
3
233