Umbra Privacy Platform
umbra.shadowproject.io
umbra.shadowproject.io
https://github.com/shadowproject/whitepapers/releases/downlo...
... that's really scary (ECDH + ECDSA + CBC mode, using OpenSSL's primitives, in an ad-hoc configuration, with plaintext compression), and also inferior to modern secure messaging schemes. People should use Signal Protocol for this stuff.
Slightly later:
The only source code I can find for this is a giant collection of C++. There's CSS referring to "ShadowChat", and a file "smessage.cpp" which roughly corresponds to the protocol sketched in that PDF. It's C++, lots of it, with a lot of memcpy.
If you want users to take your comments seriously then you should be a bit more specific about what's scare. CBC and ECDH are the basis of most modern end-to-end encrypted messaging schemes.
I agree with the key exchange, it doesn't have a proper ratchet, it's on our to-do list.
I didn't mean to come off as rude, it's a bit late and I'm quick to the trigger sometimes.
What are your concerns with our scheme? Do you have any suggestions on improving it?
We really need to implement a key ratchet, that's a sure thing.
Nobody is going to be able to sanity check this code "in real time" in response to message board comments. But I skimmed it, and I see concerns right away. For instance, it looks like the CBC IV isn't included in the MAC calculation.
In terms of primitive selections:
* CBC+HMAC is fine if you're very careful. I'd certainly do that before I tried to hand-code a real AEAD. But then, see above! A modern protocol would use either a hermetically sealed AEAD construction like GCM, or, even easier, use ChaPoly, as provided in TweetNacl.
* I can't tell what curve this is based on, but I'm guessing because this is all bitcoin related it's secp256k1. New protocols shouldn't use secp256k1, if only because it's not misuse-resistant. I'm assuming this project inherits its secp256k1 code from wherever bitcoin gets it, but anyone else who tried to interoperate with ShadowChat has to carefully avoid invalid curve attacks as well.
* ECDH is fine. Whatever. The problem is secp256k1, not ECDH.
* But ECDSA is not fine. It relies on a random nonce and explodes like a pipe bomb if so much as a single bit of the nonce is biased, which is surprisingly hard to get right. It's also difficult to make ECDSA constant time with arbitrary curves.
That's just the really basic stuff, though. Secure messaging systems are surprisingly difficult to get right. You need to understand the underlying crypto primitives very well, but then also a whole mess of almost- domain-specific stuff about messaging protocols with particular threat models.
This is not a good thing to DIY.
Later
And, I mean, I have structural concerns with this C++, too. For instance: there's a place that passes an integer computation to std::vector's resize, then memcpy's into it on the assumption that the computation didn't shrink the vector. Is this a real problem? Fuck if I know. You know what it costs per day to really audit C++ code like this?
Later Later
Sorry, I forgot: compressing plaintext creates traffic-analytic side channels. Back in 2002, people thought it was a good idea to compress plaintext, because it destroyed structure and made some attacks harder. But then CRIME was published and now we all worry about compression --- also, it turns out, the code for the exploits compression was supposed to stop gets trickier, but not impossible: see "Dancing On The Lip Of A Volcano" for more.
Also: how much of this code is your project's, and how much is inherited?
We know that thorough review does not happen instantly. I want to let you know that we have bounties and funds available for doing code review. https://shadowproject.io/en/bounties
* That's indeed true, we will be including the IV in the HMAC. Will be patched in our next version. Thank you!
* ChaPoly by TweetNaCl - dearly noted. The thing is, libs need compatibility with our curve. TweetNaCl only supports Ed25519 :/
* secp256k1 is correct.
* Do you have any references to the "exploding like a pipe bomb" problem? We're using the ECDSA from the bitcoins secp256k1 library - which is constant time and should be secure enough. Function: RecoverCompact()
We initially paid for a cryptographic review but the guy, whom we know by real name, disappeared from the planet. We're scouting for cryptographers willing to do code reviews, for a payment ofcourse.
If you're capable and interested then you should reach out to us on GitHub or Slack!
We do appreciate you doing this in your free time, so we're sending Bitcoin tips to whoever makes good suggestions. Please post your Bitcoin address if you'd like to receive some.
---
Most code is inherited from Bitcoin, custom code with cryptography: smessage.cpp, ringsig.cpp and stealth.cpp
It is OK that you don't know the best curve to use, or that an IV needs to be included in your MAC.
What is not OK is shipping privacy code to users without understanding this stuff. You don't get to learn this stuff on the job. Sorry. I know it's not fair. But to do otherwise would be even less fair to your users.
First be sure of your crypto. Then set up the bug bounty.
You can donate mine, for the MAC bug (which is severe), to Partners In Health.
Especially when issues like these: https://github.com/shadowproject/shadow/issues/25 pop up.
"There is no way to contact you on the bug bounty program"
Still, I've been around the space for a long time and I don't remember hearing too much about the Umbra project before today. That's usually a bad sign, especially given that they don't have a whitepaper.
There are a few major problems that slaughter projects like this:
1. They don't respect scalability constraints.
2. They don't actually understand what it takes to achieve privacy.
3. They don't realize that they've set up a centralized system.
4. They don't understand a lot of things about decentralization. This stuff is hard.
The only whitepaper I could find for the whole system was for ShadowChat, which says they use AES-256-CBC to encrypt the chat. AES is a symmetric encryption algorithm, and a strange choice for a chat protocol. Furthermore, CBC is not great at hiding patterns, and chat will often have a lot of patterns. Not necessarily going to cause problems, but it's another decision that I think is pretty weak, especially when there are better standards already out there.
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&c...
http://crypto.stackexchange.com/questions/25089/plaintext-bl...
As for anonymous cryptocurrency, there have been a lot of attempts at creating anonymous cryptocurrency and they are generally garbage. Either the mixing leaks information, or there is a reliance on central services, or other major problems.
Shadowcash claims to be a Proof-of-Stake cryptocurrency, which is a consensus system that has been pretty highly discredited among the more well respected researchers. The threat model is definitely different, and in general is a lot weaker both in terms of attack-resistance and in terms of decentralization that traditional Proof-of-Work. Even Ethereum has yet to come up with a Proof-of-Stake algorithm that they feel comfortable committing to, and they were so certain they'd get there that they added a feature which will break the currency unless they hardfork to PoS (at least that was the idea - given that PoS is not ready, they will probably just hardfork to a not-intentionally-broken PoW).
https://download.wpsoftware.net/bitcoin/pos.pdf https://download.wpsoftware.net/bitcoin/old-pos.pdf
A few worth taking seriously are Monero, Zcash (though Zcash depends on new/fancy/less-trusted crypto - zkSnarks), and JoinMarket. Everything else that's got academic strength behind it is too early to have a user-ready implementation.
That's just my two cents. Maybe in a few weeks there will be a whitepaper that covers the full strategies of Umbra and it ends up being pretty good. But I don't think so, given that the few decisions that I could find already suggest that this is a platform built by amateurs.
Okay.
> AES is a symmetric encryption algorithm, and a strange choice for a chat protocol.
No, it's the standard choice.
Signal uses AES-256-CBC + HMAC-SHA-256, for example.
> Furthermore, CBC is not great at hiding patterns, and chat will often have a lot of patterns.
That's not true. CBC mode isn't great at hiding patterns with IV reuse. Solution: Don't reuse IVs.
The problem with CBC mode is that it's unauthenticated, and as a result is often susceptible to padding oracle attacks that allow messages to be decrypted byte-at-a-time.
https://paragonie.com/blog/2015/05/using-encryption-and-auth...
A better mode would be GCM, or maybe GCM-SIV, or scrap AES entirely in favor of ChaCha20-Poly1305.
On the other hand I agree to switching over to a different mode to eliminate padding oracle attacks.
The IV are 16 random bytes generated by OpenSSL's RAND_bytes. https://github.com/shadowproject/shadow/blob/master/src/smes...
https://github.com/shadowproject/shadow/blob/f3fa333f8377688...
That's really hard to read.
Also, don't use OpenSSL: https://paragonie.com/blog/2016/05/how-generate-secure-rando...
Yes we're always where the critics are, they are or best source of information.
I'm very happy to see our work being reviewed. I'm not an expert cryptographer but I am capable of understanding it. (fyi I didn't code it).
Our memcmp in constant time is not the prettiest, but it's short so we roll with it :P
This project started around 2014, LibSodium was still very small back then and OpenSSL, in ours view, remains the defacto standard. Is there any particular reason on why we should move away from RAND_bytes() ?
https://github.com/ramsey/uuid/issues/80
https://github.com/nodejs/node/issues/5798
There are also some recent IACR papers (linked in the Node thread), but those are the two biggest concerns.
That link to NodeJS was a good read, I'm fairly convinced that we should ditch RAND_bytes from OpenSSL for something more secure, we'll look into LibSodium.
I've caught rumours of a possible RAND_sys_bytes which operates over the systems CSPRNG? We like to be conservative on the libs.
We'd like to tip you for your efforts, do you have a Bitcoin (or ShadowCash) address?
1D6RMsgnTAf2GLWpD91EaQvQ5ppuaZKkGTIndeed HTTPS was not enabled on the page, it was still a work in progress. It is now enabled.
We're working on an installer that will mitigate MITM issues by having releases to be signed by multiple people.
1. We do respect scalability constrains, but we decided to monitor the network and scale when needed. The idea that we have to focus our limited development time and funds on issues that are not yet a problem is a bit goofy. The idea for scaling is to include a tree like structure: 1 / \ 2 3 You split the data into numbers, based on the capacity of your node. E.G. Alice has an address of stream 2. If Bob wants to send a message to Alice he encodes the stream number as metadata into the message and connects to a node of either 1 or 2. Nodes with the same number relay messages to eachother. When Bob can't find a node with number 2 he can just share it with a node of number 1. It will finds its way down (like a stream) to 2.
Project is far from perfect, as you said, but it's worth a try. With awareness we hope that more reviews come to make this project more solid.
The whitepapers are in the number of two (chat and anon sending) : https://shadowproject.io/en/documentation
A good doc explain most of the functions in ShadowProject : https://doc.shadowproject.io/
Fully anonymous blockchain tech probably scares them.
This is probably the single most significant concept for legal privacy rights in the US.
OK, it's an altcoin scam.
Sounds like they don't even have any interesting cryptographic primitives for privacy, but they just have a mixer built into the app or something. Why would I use this over using Bitcoin with a mixer? Plus, if/when Bitcoin starts integrating ZKP/privacy cryptography like Ring signatures, it's going to be a very hard sell to use any of the miscellaneous altcoins that have some privacy-related cryptographic advancement as their only feature.
Ever heard of IND-CCA3?