Eigenwallet: Atomic Swaps in the Browser, Ethereum Swaps, Security Research and Continued Development

Who

We are the maintainers of Eigenwallet, formerly UnstoppableSwap, and have worked on XMR<>BTC atomic swaps since 2021. We know the protocol and our implementation in detail. In these years, our goal was to make this technology easy to use for everyone. We have completed three CCS proposals:

  1. XMR BTC Atomic Swaps Desktop GUI (October 2021, 52 XMR, MR 257)
  2. XMR BTC Atomic Swaps Desktop GUI - Continued development for 4 months (May 2022, 232 XMR, MR 321)
  3. From Prototype to Marketplace: Maturing the XMR-BTC Atomic Swaps Ecosystem (July 2024, 729 XMR, MR 477)

We completed the last milestone of the third proposal in August 2025. Since then we have continued the work without CCS funding.

What is Eigenwallet?

Eigenwallet is a self-custodial Monero and Bitcoin wallet with a built-in decentralised exchange. The exchange is based on atomic swaps. Two parties trade Bitcoin for Monero directly, and neither has to trust the other or a third party. The protocol uses only the Bitcoin and Monero chains. There is no token, and there is no shared pool that an attacker can drain.

Joël Gugger first described the swap in "Bitcoin–Monero Cross-chain Atomic Swap". Philipp Hoenisch and Lucas Soriano del Pino improved it in "Atomic Swaps between Bitcoin and Monero". Eigenwallet uses their protocol.

To the best of our knowledge, atomic swaps are the only working decentralised exchange mechanism for Monero that requires no arbitrators. Eigenwallet and BasicSwap are the only implementations in use today.

Statistics

Eigenwallet has settled approximately 9000 BTC of swap volume over the last year. This number comes from three sources: scrapers of the public order book, data from market makers, and on-chain heuristics. The heuristics are approximate, but they agree with the data that market makers have provided to us. The Matrix channels of Eigenwallet have more than 500 users.

The cumulative size of the offers on the Eigenwallet network from November 2024 to November 2026, with the periods without data marked

Swap volume for each month, completion time of a swap for each week and the share of successful swaps for each month, from July to October 2026

What we have worked on since the last CCS

To our knowledge, Eigenwallet is the first implementation of this protocol with large volume and with serious competition between market makers. This showed us problems that the original papers do not describe. A taker could lock the Monero of a maker at no cost. A user who lost their device during a swap could lose the funds in that swap. A swap could fail only because the two parties lost their connection. Makers compete on price, and attacks on makers became frequent and expensive. In the last year, most of our work was to make the protocol resistant to these problems.

We have merged 365 pull requests since the last CCS. Our previous CCS proposals and their monthly updates show what else we have worked on. These are some of the important changes. The list is not complete:

  • Anti-spam deposit A maker can require a deposit. If the taker refunds the swap, a part of the Bitcoin stays locked. The maker can withhold this deposit from a griefer, but the maker can never receive it. Thus a maker gets no profit from failed swaps. (#675, #899)
  • Apollo The taker writes an encrypted backup of the swap state into a Bitcoin transaction. With only the wallet seed, the user can then restore a swap in progress on a new device. (#1248)
  • Sync servers Users upload their swap state to servers, and their other devices download it from there. The servers see only ciphertext. The state is a CRDT built with Automerge, so devices merge their changes without a central coordinator. We have not released this yet.
  • Hermes The maker needs an encrypted signature from the taker to complete a swap. The taker can now send this signature in the tx_extra field of a Monero transaction. After setup, a swap can complete without a connection between the two parties. (#855, #1145, #1147)
  • Non-interactive transfer proofs The taker no longer needs the transfer proof from the maker to see that the Monero is locked. It finds the lock transaction on chain and verifies the amount with the view key. A missing proof no longer leads to a refund. (#818)
  • Wormholes Every taker with funds at stake gets a dedicated onion service to reach the maker. An attacker can flood the public address of the maker with requests. This no longer stops swaps that are in progress. (#944, #955, #1228)
  • Two critical vulnerabilities We found and corrected two critical vulnerabilities before an attacker could use them. The security research section below gives more details. (#1243, #1165)
  • Protection against transaction pinning In a pinning attack, one party spams the Bitcoin blockchain to block the transaction of the other party until a timelock expires. The cancel and refund transactions are now version 3 (TRUC) transactions, and the fees of all pre-signed transactions follow one agreed fee rate. (#1243)
  • Lock input ownership proofs A taker must prove control of the Bitcoin inputs before the maker signs anything. This increases the cost of a denial-of-service attack during swap setup. (#1243)
  • Reserve proofs A maker can attach a Monero reserve proof to a quote. A taker can then examine the liquidity of the maker before the taker commits funds. (#809, #812)
  • Maker discovery We wrote new code for how a taker finds makers and connects to them again. Rendezvous points now register with each other, and takers can reach makers through websockets and relays. (#645, #692, #698, #717, #947, #1125, #1198, #1218, #1222)

Proposal

We ask the community to fund six months of work on Eigenwallet. The scope is fixed. Four pieces of work have their own milestones: the Firefox extension, two refactors of the maker and Ethereum swaps. The other work is security research and work on the protocol and the GUI. It is paid by the hour, and each of the two of us does 15 hours of it per week.

1. Firefox extension (200 XMR)

Today a swap requires our desktop application. A vulnerability in it can give an attacker access to the full system. It must include its own copy of Tor, and it is difficult to install on temporary systems such as Tails.

We will make the swap client available as a Firefox extension. The user installs it with one click and does not download a separate application.

The extension will work like MetaMask. A site can ask the extension to start a swap, and the user confirms it in the extension.

A market maker can add a swap button to a site of their own. Users then start an atomic swap with that maker through the Eigenwallet extension. The maker can stay anonymous and advertise the site on forums or in other places. An instant exchanger can do the same with a maker of its own. The community can also maintain sites that list reliable makers.

A maker cannot steal funds, but a maker can force a refund and waste the time and fees of the user. It is not practical to build a reputation system into the protocol. Reputation must come from the community instead: from community sites, users and market makers with their own site.

A browser extension can also be more secure than a desktop application. It runs in the Firefox sandbox, so it has much less access to the system. In Tor Browser, the user also gets the network anonymity of Tor. The build of the extension is reproducible on all architectures, so anyone can verify that the released extension matches the source code.

The swap client is written in Rust and compiles to WebAssembly. A browser is a more difficult environment than a desktop application. A swap waits for minutes or hours between steps, and the browser can stop a background worker at any time. If the user deletes the browser profile, the browser also deletes the wallet. We built the sync servers, Apollo and Hermes to keep swaps safe in these conditions.

The extension connects to a set of remote Monero nodes and a set of remote Electrum servers, and it reaches the maker through a Tor circuit. The connections to the Monero nodes and the Electrum servers also go through Tor and are anonymous. The maker runs its own Monero node and Electrum server.

The WebAssembly swap engine is portable. It can run on all systems, and other wallets can embed it easily. For example, it can become an extension for Feather Wallet or a part of Cake Wallet.

The milestone is complete when we have released the extension and a user can complete a swap on mainnet with it.

A site asks for a swap and the Eigenwallet extension shows a popup to confirm it

Eigenwallet in Tor Browser connects to remote Monero nodes, remote Electrum servers and, through a Tor circuit, to a maker

2. Security research (57.35 XMR)

In the last year we found and corrected two security issues before an attacker could use them. The first is a fee race between cancellation and refund near the first timelock (#1243). A dishonest taker could use it to get the Bitcoin and the Monero. It was in the baseline protocol for years, and BasicSwap had the same type of problem. The second is a bug in swap setup that let a dishonest taker burn funds (#1165).

An attacker used a third issue before we found it. In May 2026, a dishonest taker burned their own Bitcoin to burn the Monero of two makers. The attacker could not steal funds, and takers were not at risk. We immediately asked almost all makers to stop their maker software, and we released a fix in less than 24 hours (#1051). The post mortem gives more details.

We will search for security issues in the protocol and in our implementation. It is possible that there are more vulnerabilities, especially denial-of-service vectors in the peer-to-peer layer. Unknown parties have already attacked makers with very large DDoS attacks, and we have built countermeasures.

A large part of this work is manual. For example, we will write tools such as fuzzing engines by hand. We do the other part of this work with large language models. Our Defense Fund will pay for the tokens.

3. Protocol and GUI (38.23 XMR)

We will continue to improve the protocol and the desktop application. These are the items we plan to work on first:

  • Support the FCMP++ and Carrot upgrade of Monero in the wallet and in the swap protocol.
  • Let the maker increase the fee of the Bitcoin redeem transaction with pre-signed fee tiers (#984).
  • Send the key for the recovery of Monero after a punish through a Hermes message (#1192). A punished taker then receives the Monero without a direct connection to the maker.
  • Use Merkle proofs to verify the confirmations that the protocol relies on.
  • Move the Bitcoin wallet from the deprecated bdk library to bdk_wallet (#1171).
  • Let a taker do more than one swap at the same time (#1262).

We will also continue to correct bugs, extend the integration tests, answer issues and publish releases. We will review pull requests from external contributors where possible. We must limit this review, because a very large number of the pull requests that we receive are low-quality output of large language models.

4. P2P microservice (15 XMR)

Today libp2p and Tor run in the same process as the wallets and the swaps. This process accepts connections from anyone. A vulnerability in the network code is thus a direct threat to the part that holds the funds. We will move libp2p and Tor into a separate service. If an attacker gets control of that service, or if the service crashes, the wallets stay safe and the swaps in progress continue (#1249).

Today the network code and the wallets share one process. After the change the network code runs in its own service

5. Move to monero-oxide (15 XMR)

Today the maker uses the C++ wallet code of Monero through a bridge between Rust and C++. This bridge is complex and decreases the memory safety that Rust gives. We will replace the C++ wallet code in the maker with monero-oxide, a Rust implementation. monero-oxide was built for Serai and has been audited. The change makes the maker simpler and decreases its attack surface (#1250).

Today the maker calls the C++ wallet code through a bridge. After the change the maker uses monero-oxide directly

6. Ethereum to Monero atomic swaps (200 XMR)

We will add atomic swaps from Ethereum to Monero to Eigenwallet. A user who holds ETH swaps it for Monero with a maker, in the same application and with the same steps as a Bitcoin swap today.

The protocol exists already. Elizabeth Binks described it in "Cross-chain Atomic Swaps between Ethereum and Monero", a paper the Monero community funded. The taker locks ETH in a smart contract and the maker locks Monero. When the maker claims the ETH, the contract reveals the secret that gives the Monero to the taker. If the swap fails, the taker's refund reveals the secret that returns the Monero to the maker.

The structure is the same as in our Bitcoin swap, but a smart contract replaces the pre-signed Bitcoin transactions.

The swap contract is not specific to ETH. After this work, it is easy to add ERC-20 tokens on Ethereum, such as DAI or USDT, and other EVM chains.

The steps of an Ethereum to Monero swap between the taker, the swap contract, the maker and Monero

For Ethereum swaps the extension also connects to remote Ethereum nodes, and the swap contract holds the ETH

Two projects implement this protocol today: AthanorLabs/atomic-swap, which was the first and is now archived, and MoneroSwap. Both teams have done good work.

The swap itself is not the difficult part. The difficult part is everything around it: recovery after a lost device or a lost connection, protection against griefing and denial of service, the discovery of makers, and an application that a new user can operate. We have solved these problems for Bitcoin swaps, and the solutions have run in production for years.

An Ethereum swap is very similar to a Bitcoin swap. It makes more sense to add Ethereum to this tested machinery than to bring the code of another project into ours. Ethereum swaps then get the same protections from the first release, for example the deposit that a maker can withhold from a griefer.

The second reason is the users. Eigenwallet has approximately 500 users in its Matrix channel, several hundred BTC of swap volume each month and a competitive market of makers. Neither of the other projects has many users yet. Ethereum swaps in Eigenwallet get these users and makers from the first day. It is not feasible to move them to a separate project, because users and makers do not put significant funds into software that does not have a long record of use.

Market makers from the community have told us that they will supply more than 500 ETH of liquidity after the launch.

Funding

We request 525.58 XMR in total.

  • 95.58 XMR are for 720 hours of security research and protocol work in six months. Each of the two of us works 15 hours per week on this. It is paid as 15.93 XMR for each monthly milestone of 120 hours. We request a payout only after we complete the hours and post the monthly update.
  • 200 XMR are the milestone for the launch of the Firefox extension on mainnet.
  • 30 XMR are the two refactor milestones of 15 XMR each.
  • 200 XMR are the milestone for Ethereum to Monero swaps.

Our hourly rate is 60 EUR. We calculated the XMR amount of the hourly work with an exchange rate of 452 EUR/XMR, the price on Kraken on November 16, 2026.

Area Hours Amount
Firefox extension: launch on mainnet 200 XMR
Security research 432 57.35 XMR
Protocol and desktop GUI 288 38.23 XMR
Maker peer-to-peer stack in its own service 15 XMR
Maker Monero wallet on pure monero-oxide 15 XMR
Ethereum to Monero atomic swaps 200 XMR
Total 720 525.58 XMR

We will post a detailed update in the comment section every month, as we did for our previous proposals.

This proposal expires on August 31, 2027.

Edited by binarybaron

Merge request reports

Loading
Loading