Mavis

[Sebastien Guillemot @SebastienGllmt]

Happy to have shared at the Solana Summit today our new Solana wallet support for Midnight!

@solana users will be able to do private swaps and interact with private DeFi directly from their Solana wallets (no separate install required!)

[Sebastien Guillemot @SebastienGllmt]

📢Calling all builders

Contracts are LIVE on Midnight mainnet

[Liquid AI @liquidai]

> Decision models page: https://t.co/5M2jFbC9yP
> Migration guide: https://t.co/mN5qh9PrJv
> Demo tutorial: https://t.co/35TL4FueMl

Enjoy 😊

[Matteo @0xMetamatt]

After many years of work, CIP113 has been officially merged to the main branch of Cardano CIPs.

This means the official standard for programmable tokens on Cardano, including securities, is out.

Many great minds have contributed to the standard, but in particular @CryptoJoe101 has worked very hard to complete it and make it production ready.

Now the standard is in the hands of the old and new Cardano projects to use!

[Mavis @mavistran21]

@theo I completely agree with you. Jev is not suited for jobs that involve speculating on complex data without reasoning.

[YUTA-Cardano/CPA(DMは全て詐欺) @yutazzz]

NIGHT償還について、ミッドナイト公式サイトから進めても、何故かエラーでうまくいかない方へ

色々やって償還させる機構を組みました、先ほどサイトでうまくいかなかった方も償還通せました、よろしければお試しください🙏

これでも、うまくいかない場合、ヘルプボタンをつけたのでそちら押していただくと、デバッグデータ私に届きますので助かります🙏

https://t.co/PD6J4Xrc6U

[Input Output Group @IOGroup]

@blockpqr @hegaleon Read about the move: https://t.co/XF5CWqpJmz

[jhey ʕ•ᴥ•ʔ @jh3yy]

spritesheets are, and always will be, a fun trick! ⚡

translate with css based on pointer position

grace note™ blink included ♪

[Intersect @IntersectMBO]

DReps and SPOs: 7 days remaining ⏳

Cardano governance is driven by DReps, Stake Pool Operators, and Constitutional Committee, with different actions requiring different voters.

Two actions expiring September 1 require separate approval from both DReps and SPOs:

🔹 Update Constitutional Committee 2026
🔹 minPoolCost and Plutus memory limits update

For these actions, DRep >67% and an SPO threshold >51% is required - documentation available to support SPO voting can be found 👇️

https://t.co/xE3rou5NYD

[Chiro Visuals @chirovisuals]

I've extended my open source VFX playground built with @threejs. New 10 fully real-time customizable abilities. 3 Radial abilities, 3 boosts, 3 portals & 1 Linear casted ability.

For this project I used Minimax-M3, one of the cheapest models around & impressive for what it costs.

Everything you see is generated: no textures, no sprite sheets, no baked meshes.hand-written GLSL, GPU-simulated instanced particles, world position resolved in the vertex shaders...

Source code : https://t.co/dNKjM2hviN

[Iron Bramble 🪻 @IronBramble]

[1/2] Some of you might recognise the procedural animation we did in our last game - Rusted Moss. We are using the same techniques for Iron Bramble. This time, with dress physics!
#indiedev #indiegame

[比特币橙子Trader @oragnes]

人类历史上首颗去中心化的链上CPU诞生了，有意思！

有人在 BNB Chain 上造了一颗 2.22Hz 的 CPU，虽然慢得离谱，但这个实验比我想象中有意思，于是抓紧跟AI研究了下，发现越研究越上头，怎么办？😂

@Blonskr 真把一颗 4 位处理器做到了链上。

一个 Token 对应一颗“晶体管”，再用与非门、触发器把电路拼起来，整个处理器由 2251 个逻辑单元构成，对标 1971 年的 Intel 4004，已经完整跑通了 1、1、2、3、5、8、13 的斐波那契数列。

最魔幻的是它的时钟。普通 CPU 靠硬件时钟疯狂跳动，这颗 CPU 靠 BNB Chain 出块：

0.45 秒出一个区块，它才往前算一个周期，所以主频只有 2.22Hz。

假设拿 5GHz 的现代 CPU 对比，两者差了大约 22.5 亿倍。

真想靠缩短出块追平，就得做到 0.2 纳秒一个区块，光在这段时间里也只能跑大约 6 厘米。

所以它基本不可能拿来替代电脑里的 CPU。但我觉得这东西最牛逼的地方也不在性能，而是把 CPU 的“时钟”从一块物理晶振，换成了整个区块链的共识。

CPU 本来就是状态机，区块链本质上也是一台全网共同维护的状态机，现在有人真的把一台处理器塞进了另一台巨型状态机里。

于是出现了一种很奇特的计算机：

跑得可能是全世界最慢的，但没有一个人拥有它的关机键，全网都可以验证它每一步怎么算出来。

下周作者准备把协议完整开源，还能像拼乐高一样继续组合更复杂的链上处理器。

这种东西短期没什么实用价值，甚至很像行为艺术。

但 Crypto 最有意思的实验，很多时候就是先造出一个以前根本不该存在的东西，再慢慢研究它到底有什么用。

[Atlas @crptAtlas]

https://t.co/Gs2bPxxlTD

[Santiago Carmuega @santicarmuega]

I wanted to learn more about Leios, so I vibe-coded this visualization TUI on top of Pallas.

It's not meant for operations, just a personal visual aid to understand the flow of the protocol. It was quite useful, but it also raised some new questions.

What you can see is an initiator node connected to the Musashi Dojo. The TUI is divided into two horizontal lanes:

- top lane: the Praos part, a sequence of ranking blocks (RBs)
- bottom lane: the Leios part, a sequence of endorser blocks (EBs)

The X-axis represents time, so you see blocks ordered from left to right, representing the sequence of the chain.

The interesting part to notice from the video: the heavy lifting of Leios (votes, tx processing, etc.) happens between RBs, making use of the idle time.

Being a vibe-coded PoC, the TUI uses some heuristics to work around much of the complexity and the current protocol gaps. For example:

- the link between RBs and EBs is estimated using slot proximity
- the votes are just a counter, and the denominator is the peak number of voters seen so far

One question that popped up and I'm still looking into:

the idle time doesn't seem to be enough to observe all the EB votes and download all the txs. This is probably related to data diffusion; maybe client prioritization on the responder side; maybe just bad performance on my code. 🤷

[FluidTokens @FluidTokens]

As a protocol building on Cardano, are you ready to join BIFROST?

Are you ready to get Bitcoin liquidity on your dApp?

Are you ready to potentially onboard a completely different cohort of users?

Even if you think you are not, you actually are:
Our Dev team will assist you for FREE. 😤

You can find more information on how to apply here:

https://t.co/L9WTslOsQL

[Mavis @mavistran21]

@FluidTokens Interesting

[Intersect @IntersectMBO]

Readiness resources 👇

Upgrade readiness pages - https://t.co/Ja3kCCMNSx
Cardano constitution - https://t.co/hFqNuJSM5o

SPO readiness:
PoolTool - https://t.co/qm3gZDEfJY
Cexplorer - https://t.co/rZB09yZTtY
ADAtool - https://t.co/jeFOftSycU

Exchange readiness:
CardanoScan - https://t.co/SfVrpFar78
Cexplorer - https://t.co/Z5XtcsBGSO

[Input Output Group @IOGroup]

1/ Blockfrost belongs to the people who build on it.

Today, the proposal to make it a free, community-governed access layer, owned by the ecosystem, is live on-chain.

[Jonathan @solidsnakedev]

The signing math

When you sign a message with Ed25519, two values get published on-chain as your signature: R and s.

They're computed like this:
- R = k · G (k = the secret nonce, G = a fixed public point)
- s = k + H(R, A, M) · a

Where:
- k = the nonce (must be secret)
- a = your private key (the thing that must never leak)
- A = your public key
- M = the message (transaction)
- H(...) = a hash anyone can compute from public data

(Simplified, the real scheme works modulo the curve order and the nonce is derived from a secret value inside the seed, not the raw seed. But the shape below is exactly what matters.)

Both R and s are public. They're literally in the signature on the blockchain.

Now solve for the private key

Look at the second equation and rearrange it:

a = (s − k) / H(R, A, M)

Everything on the right side is public except k, the nonce. So the only thing protecting your private key a is that nobody knows k.

That's the whole game. Secrecy of k = secrecy of your key.

What SecondFi broke

Correct Ed25519 derives the nonce from a secret:

k = H(secret_seed, M) ← depends on something private

You can't recompute k, because you don't have secret_seed.

SecondFi instead did something equivalent to:

k = H(M) ← depends ONLY on the public message

Now anyone reading the chain can:

1. Read the transaction M → compute k = H(M)
2. Read s and R from the signature
3. Compute H(R, A, M)
4. Plug into a = (s − k) / H(R, A, M) → recover the private key
A toy numeric walkthrough

Forget elliptic curves; use plain arithmetic to see the shape.
Say:

- Private key a = 7 (secret)
- Message hash H(R, A, M) = 3 (public)
- Nonce k = 5

Then the published signature value is:
s = k + H·a = 5 + 3·7 = 26

Good system: an attacker sees s = 26 and H = 3, but not k. Two unknowns (k and a), one equation unsolvable.

SecondFi's system:
the attacker recomputes k = 5 from public data.

Now:
26 = 5 + 3·a → a = (26 − 5) / 3 = 7

The private key falls right out. One signature, no guessing.

Why this beats classic "nonce reuse"

The famous bug, the one that broke the Sony PlayStation 3 and some early Bitcoin wallets, was reusing the same k across two different messages. That took two signatures to cancel the unknowns and solve.

Here you need only one, because k isn't unknown at all. It's computable from data already on chain. Every single transaction a user ever signed is a standalone leak of their key.

The fix was never "understand the curve math." It was "don't be the one writing it."

Libsodium and noble/ed25519 are good examples: both open, both audited, and both give sign() no parameter to supply the nonce.

[Mavis @mavistran21]

@ItsDave_ADA Hi @ItsDave_ADA , Do you have the addresses of the CEXs on Cardano? I also have a tool that can trace them.