3/ The part I am more interested on: darkfid logs a verifying key for every circuit on startup. The recent testnet reset fixed a bug where two proofs were linked through Poseidon's preimage resistance instead of an explicit constraint. That's a bug class I want to read next.
I put a DarkFi testnet node + miner on a Raspberry Pi 5. Runs 24/7, reachable over self-hosted WireGuard, data on NVMe.
Wrote up some thoughts and notes on the process:
https://t.co/pZLAsKCtUY
2/ The useful part for anyone on ARM: a minimal Pi OS is missing half of what DarkFi's build wants. wasm32 target, cmake for RandomX, ALSA, sqlcipher. Install them up front and skip the failure loop.
I followed old Anchor tutorials with modern Anchor 1.0 documenting the concrete diffs I found: project structure, instruction serialization, ctx.bumps, token_interface, and why PDAs stayed invariant.
Plus: LiteSVM vs the 3-terminal testing setup.
๐งต๐
https://t.co/FRuhoBZE3Q
Still have challenges 7 & 8 left (AES ECB) โ follow-up when I get there.
All implementations in a Rust workspace with a reusable crypto_utils crate. Diagrams in the post show the bit-level operations step by step.
New post: I implemented the first 6 Cryptopals challenges from scratch in Rust โ hex codecs, base64, XOR, frequency analysis, Hamming distance, and a full repeating-key XOR breaker. Without libraries.
Notes, diagrams, and code:
https://t.co/Mr8W6k1qWL
Takeaway: breaking repeating-key XOR decomposes into N independent single-byte XOR attacks once you guess the key length via normalized Hamming distance.
Hard problem โ many easy problems. Same pattern everywhere in crypto.
Last week I got targeted by a supply chain attack during a fake Web3 job interview.
npm install โ 44 minutes of silent RCE on my machine.
I set up an isolated Hetzner VM, reproduced the attack, captured every packet, and reverse-engineered the full C2 protocol.
Full breakdown ๐งต