Fellowship Notes
A weekly log from inside the protocol fellowship — what I’m reading, building, and getting wrong. Written in Markdown, published every Sunday.
pq-sap: Final Report
The closing report for pq-sap, a post-quantum stealth address scheme for ERC-5564: five deliverables shipped, one ERC draft left deliberately unfrozen, and an honest account of the pivots, the isolation, and where the scheme goes next.
Read the report →What a journey it's been
A quiet two weeks of wrap-up: every ERC-8441 review comment addressed, the Lean 4 formal verification handed to the author with ML-KEM anonymity backed by ePrint 2022/1696, a DefCon 33 PQC talk worth watching, the VCVio issue closed, and the pq-sap repo cleaned down to a journal.
ERC-8441: Serendipity in BAAL
A couple of weeks before Devcon in Mumbai, my parallel PQ stealth-address effort lands as ERC-8441 (BAAL, am I too 1337 to see it?): ML-KEM-768 + secp256k1 ECDH behind one payment secret. I left ten review comments, proposed generic KEM recipient-anonymity games in VCVio, watched Vitalik lean into hash-only and fresh addresses, and queued up daisugi.fyi.
Inception
A more autonomous research setup while working and travelling, two draft post-quantum stealth-address schemes, and an encouraging parallel effort that validates keeping ML-KEM key exchange separate from the moving spending layer.
Framemogging and Demoing
Frame transactions got scheduled for inclusion and now run on a Kurtosis devnet with a live pq-sap demo, EIP-8304 got declined, native UTXOs open a nicer derivation story, the gas benchmarks say ~$200 per signature verification — and EIP-8288 mempool aggregation is the only way out.
X-Wing, Frames, ecrecover, SPHINCS⁻ and VCVio
Wrapping the proposal up for an ethresear.ch thread — Circle's ecrecover trick for hybrid signatures instead of ERC-7913, office hours #11, two VCVio PRs from de-slopping machine-written Lean 4, why hash-based key exchange doesn't survive Grover, EIP-8304 UTXO discovery, Nethermind's frame-transaction prototypes and the 8130/8141 drama, SPHINCS⁻ as a ~77× cheaper spending option, and a demo app.
Bytes in, leverage out
After the proposal presentation — the published pq-sap repo with ML-KEM/ML-DSA scripts and vectors, how I plan to use AI during the cohort, ERC-7913 signer abstraction for spending, ERC-8373 for pre-Q/post-Q key binding, and getting comfortable with the VCVio oracle model.
Trees, hashes and Lean
Researching VCVio and Lean 4 for the security-analysis part of the proposal, splitting key exchange from signatures, and a detour into Merkle-based alternatives for the stealth meta-address registry — with a Poseidon note that did not age well.
Back to the lattice
Returning to post-quantum stealth addresses — splitting key exchange from spending, an ML-KEM announcement PoC validated live on Sepolia at 67,580 gas, and a first stab at formally verifying the algebra in Lean 4.
Marrying the proposals
Two weeks of drafting the EIP-8304 project proposal with Vansh — Fable as a first pass, reading the testing tooling, splitting the work, and realizing the EIP also unlocks cheap temporary verifiable state.
Scoping 8304
Going deep on the EIP-8304 trustless log indexing spec and shaping a first project proposal.
SSZfication
Reading through the ton of specs and SSZ related work that is currently ongoing on the EL.
Ð-day
First week of the EPF — execution layer, good first issues, and post-quantum stealth addresses before settling on a direction.