Announcing the release of YamnOfflineTools-NG, a standalone, pure-Go
offline utility suite for constructing YAMN v2 mixnet packets.
Announcing the release of YamnOfflineTools-NG, a standalone, pure-Go
offline utility suite for constructing YAMN v2 mixnet packets.
Gabx wrote:
Announcing the release of YamnOfflineTools-NG, a standalone, pure-GoThanks for releasing it.
offline utility suite for constructing YAMN v2 mixnet packets.
Building packets offline and moving them by QR code is a nice idea.
I compared yp/yp.go with crooks/yamn v0.2.6 (client.go and packet.go).
The packet code is taken from there, with the same logic.
That covers slotFinal, slotData, encodeHeader, encMessage,
slotIntermediate and their methods.
It also covers encodeMsg, seqIV, aesCtr and mixprep.
The changes are small:
panics became error messages with os.Exit(1), and Pubring became PubRing. blake2s.New became New256, which gives the same output.
The reuse is fine but crooks/yamn is MIT licensed
("Copyright (c) 2014 Steve Crook").
MIT requires that notice to be included with copies
or substantial portions.
The repo has no LICENSE file right now.
The README credits only Ch1ffr3punk and says the suite was
"redesigned from the ground up".
Suggested fix:
- Add crooks/yamn's MIT notice to the repo, e.g. as
LICENSES/MIT-yamn.txt or in a NOTICE file.
- Say in the yp.go header and in the README that the packet construction
is derived from github.com/crooks/yamn (MIT).
- Add the AGPL-3.0 text as LICENSE, since the README names AGPL-3.0 but
the file is missing.
MIT code can be part of an AGPL project, so only the notice is needed. Nothing else has to change.
A small unrelated note:
go.mod says "go 1.27", while the guide says Go 1.25 or newer.
Since you build on crooks/yamn:
we maintain a fork of it at https://github.com/go-while/yamn.
crooks has been inactive for over a year,
so we went through the code in two reviews and fixed what we found.
Stock and fork remailers can run side by side in the same network.
What it does differently:
- A malformed packet or file no longer crashes the remailer.
- A middle remailer only forwards to remailers in its keyring, without
line breaks in the address, so a sender can't use it to mail arbitrary
addresses or inject headers.
- Messages larger than one packet are delivered.
Stock clients reuse one packet ID for all chunks, so exits drop chunks
2 and up.
- SMTP to a configured relay verifies the certificate, with an option to
require TLS.
SMTP connections have timeouts.
- The replay window is closed: ID log entries last longer than the
maximum packet age.
- A failed keyring download no longer overwrites a good keyring.
- The client can merge keyrings and stats from several pingers, with a
quorum on keys.
- It can verify signed pinger files (minisign).
- Key pinning: a remailer key that changes earlier than a normal
rotation is held for 24 hours and logged before use.
- Tor support, including .onion relays.
- yamn refuses to run as root.
There is also go-echolot (https://github.com/go-while/go-echolot),
a pinger in Go meant to replace the Perl Echolot.
It signs its output, publishes a new remailer key only after a ping
through that key comes back, and holds unexpected key changes for 24
hours. One instance has been pinging the public network since late October.
Testing isn't finished yet.
We run the fork on an 18-node private test network with three go-echolot pingers.
Early and lost key changes pass automatically, on the pingers and on the client.
Still open: a natural key rotation of all nodes, a mixed run with stock
yamn nodes, and the long-term test.
Until that's done, I wouldn't call the fork production-ready.
Feedback from an operator would be welcome.
- Hardening against malformed packet panics, enforcing non-root
execution, and adding native Tor/.onion transport bring the remailer
codebase up to modern operational security standards.
The real defense-in-depth is pairing YAMN with modern mixnets like the
Nym Network (Sphinx packets, Poisson delays, and loop cover traffic) to achieve genuine network-level metadata privacy against a global adversary.
5. Collaboration and Testing
I would be very glad to collaborate. I am interested in testing YamnOfflineTools-NG against your 18-node test network and exchanging operational feedback on go-while/yamn and go-echolot.
Let me know the best channel to coordinate (via this thread, email, or
git tickets). Once again, thank you for the sharp peer review and for revitalizing the YAMN mixnet ecosystem.
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 57:40:05 |
| Calls: | 1,265 |
| Calls today: | 1 |
| Files: | 1,362 |
| D/L today: |
4 files (9,632K bytes) |
| Messages: | 285,926 |