AI + human hybrid security systems for the cyborg era. Trusted by @Uniswap, @YearnFi, @yAuditDAO and more!
- Tweets 230
- Following 2
- Followers 1,236
- Likes 239
Pinned Tweet
Uniswap Labs is adding a new layer to its security stack: Zero Cool
Going forward, Uniswap Labs will run Zero Cool Continuous Coverage across all active smart contract repos.
It’s been amazing working with this forward-thinking team throughout the rigorous evaluation process.
We’ll be sharing more details in an upcoming blog post. For now, we’re excited to welcome Uniswap Labs as the newest Zero Cool enterprise customer!
24
8
100
11,523
Zero Cool retweeted
There’s a lot happening in AI, and a lot happening in security.
At @yAuditDAO, we’re already bringing those two worlds together: @ZeroCool_AI is part of the stack we use in our audits, at no extra cost to clients.
This Monday, we kick off yBorg, our AI + Auditors fellowship, with a selected cohort — and we’ll be sharing part of the journey publicly.
We are excited to announce that Zero Cool finished 1st 🥇 in the @CRX_FX engagement on @sherlockdefi's recent Audit Engine competition!
4
22
1,657
Zero Cool retweeted
Crazy how you can get these results in a few hours for codebases that seem nearly I accessible for the average auditor. Good seeing more of these contests come up.
We are excited to announce that Zero Cool finished 1st 🥇 in the @CRX_FX engagement on @sherlockdefi's recent Audit Engine competition!
1
15
963
We are excited to announce that Zero Cool finished 1st 🥇 in the @CRX_FX engagement on @sherlockdefi's recent Audit Engine competition!
The @CRX_FX team completed an AI-only Sherlock Audit Engine run on their zkVM Risk Engine + FX Clearing Core.
CRXFX is bringing FX rate locks on-chain: 30+ corridors, custom contracts, and instant USDC settlement.
3 AI auditors: @v12sec / @zerocool_ai / @savantchat
51 submissions
21 accepted findings
1 High | 8 Medium | 3 Low | 9 Info
8
2
52
6,651
Biggest takeaway: every AI agent in this engagement found something unique.
Running multiple AI auditors on one codebase gives protocols better coverage, since each has its own strengths.
Congrats to @CRX_FX, @v12sec, and @savantchat. Thanks to @sherlockdefi for running it 🤝
1
7
368
Want to learn more about Zero Cool?
If you're a security researcher or developer interested in using Zero Cool to secure your protocol, reach out now to request a demo (DM is open!) or visit us at zerocool.ai.
4
163
Biggest takeaway: every AI agent in this engagement found something unique.
Running multiple AI auditors on one codebase gives protocols better coverage, since each has its own strengths.
Congrats to @CRX_FX, @v12sec, and @savantchat. Thanks to @sherlockdefi for running it 🤝
1
2
37
Want to learn more about Zero Cool?
If you're a security researcher or developer interested in using Zero Cool to secure your protocol, reach out now to request a demo (DM is open!) or visit us at zerocool.ai.
1
17
If your AI security agent only reads production code and ignores the tests, it is probably leaving valuable security signals on the table.
Tests tell you a lot about what the developers expect the system to do, i.e., valid states, trusted actors, assertions, expected behavior, invariants, and so on. However, tests can also give you a false sense of coverage.
Sometimes, it can miss out on an edge case that quietly gives an attacker privileged access, an entire class of inputs, or an assertion around a critical state transition.
There is almost always some alpha hiding in the test folder, and that is why we built a Zero Skill for digging through it: Test Terminator.
It reads the test suite, picks out the security-relevant assumptions, traces them back to the production code, and checks whether they actually lead to a security failure before reporting a finding.
The skill has already contributed to 84 verified high and medium findings (10C/36H/38M) in Q1 and Q2, and we highly recommend it. See more details in the write-up below.
nitter.net/ZeroCool_AI/status/208…
10
799
