1 Hour with @PatrickAlphaC where I cover:
1⃣ how I break down stateful fuzz testing by invariant types and contract lifecycle
2⃣ my favorite general heuristics which I use to find all sorts of bugs in many different codebases
3⃣ mindset and ultimate recipe for success
Link👇
It's been an ongoing pleasure to work with our expert @cyfrin audit team led by @DevDacian and @0xStalin.
Truly no stone left unturned as we put security at the forefront of our priority list in developing @LotusFi_
gpt-6.2-sol unusable as bad as opus 5 which was a horrible release.
very pedantic & manic, makes a mountain out of a molehill.
disappointing release, switching back to gpt-6.1-sol for external cross-model review.
👑 opus-5.5 still king, anthropic on the throne 👑
Using gpt-6.2-sol high as external reviewer, hit multiple cyber safety refusals on same work which gpt-6.1-sol never refused.
Automatic fallback to gpt-6-sol with daybreak blue enabled worked, but an unpleasant surprise that openai tightened default cyber restrictions.
Also gpt astra/sol refused to implement my cyber safety refusal automatic fallback mechanism but opus-5.5 high implemented it with no complaints & it is working great, cheers anthropic :-)
the dark forest is everywhere now. we are heading into a world where any non-trivial app (let alone complex smart contract protocols) will effectively require a monthly security spend or be pwned
The main takeaway from these exploits is that as we have always argued, the only reason most stuff wasn’t hacked was the cost of finding the attack (the EV for an hacker was negative).
AI changed this.
Huge vindication for crypto.
This week running 6 private audits in parallel (7 total) engaging 21 top auditors!
Great mix of EVM, Solana, Stellar & 2 engagements include Formal Verification!
Really 🔥 protocols being audited, our elite auditors pushing as always giving their all securing our clients 🚀
At Lotus, security is our number one priority.
Our pillars: economic design, smart contracts and op-sec.
We’re publishing an Economic Risk Report from @Cyfrin on how Lotus Protocol allocates capital, interest and losses across our connected markets.
lotuslabs.net/research/cyfri…
To confidently allocate capital onchain, allocators need a high level of assurance and due diligence.
The bare minimum is security: is this vehicle secure enough to allocate to?
Cyfrin Security has a proven track record of helping teams improve security posture.
Next, risk: how do the protocol and its assets behave under stress, who absorbs losses, and how does the risk spread?
Cyfrin Risk now helps teams answer that.
Our Economic Risk Report for @LotusFi_ examines how Lotus Protocol allocates capital, interest, and losses across its connected markets.
At Lotus, security is our number one priority.
Our pillars: economic design, smart contracts and op-sec.
We’re publishing an Economic Risk Report from @Cyfrin on how Lotus Protocol allocates capital, interest and losses across our connected markets.
lotuslabs.net/research/cyfri…
At Lotus, security is our number one priority.
Our pillars: economic design, smart contracts and op-sec.
We’re publishing an Economic Risk Report from @Cyfrin on how Lotus Protocol allocates capital, interest and losses across our connected markets.
lotuslabs.net/research/cyfri…
This week @cyfrin running 7 private audits in parallel (9 total) engaging 21 top auditors!
Been an insane bull market year for private audits, blessed to have grown our business substantially while working harder than ever to help keep our clients safe 🫡
Top 3 Subscription Models
🥇 Opus-5.5 High: amazing for dev & auditing, gives tons of usage for $200/m
🥈 Fable-5.1 & Astra High-Max: use sparingly for planning/consults; great but eats usage fast
🥉 Sol-6.1 High: best mid-tier model, good with near-infinite usage for $200/m
Introducing a new model that 99.99% of you plebs have no access to, haha suckas!
One day we'll expand access to you once it isnt frontier anymore, enjoy being in the permanent underclass roflmao :-P
Introducing GPT‑6 Cyber: a major advance in security research that pairs stronger capabilities with better model alignment.
The model powered the Linux kernel research we announced yesterday, and security teams at companies like Mozilla, Palo Alto Networks, Calif, and Cisco—alongside our own at OpenAI—are already putting it to work.
GPT-6 Cyber is available to select Daybreak Red customers and we are expanding access over time.
openai.com/index/introducing…
Great to see our partners @Firelightfi and @LidoFinance teaming up, with Firelight cover now coming to Lido Earn.
As a member of the Firelight Risk Consortium, we’re glad to be part of building stronger protection for onchain assets.
Firelight is bringing onchain cover to @LidoFinance Earn, protecting Lido's EarnETH and EarnUSD vaults against onchain risks.
Terms and capacity are onchain and are reassessed every 30 days against each vault’s risk.
→ app.firelight.finance/
When Opus-5.5 came out I used it to re-write existing workflows/loops/harnesses/etc from scratch using everything I'd learned plus new ideas, aiming for lean lightweight performance.
The results since have been PHENOMENAL.