Vauban retweeted
Post-quantum accounts already pay on Starknet mainnet.
A Falcon card spent a private #STRK20 note this week.
PQ-signed. STARK-proved.
We don't stop at @Starknet: @base ⇄ Starknet, both ways, mainnet.
Testnet: one command. pay.vauban.tech
voyager.online/tx/0x1c30c6d9b… #PQS
A Falcon card spent a private #STRK20 note this week.
PQ-signed. STARK-proved.
We don't stop at @Starknet: @base ⇄ Starknet, both ways, mainnet.
Testnet: one command. pay.vauban.tech
voyager.online/tx/0x1c30c6d9b… #PQS
3
2
3
75
Vauban retweeted
Bridged from @base, spent in private on @Starknet mainnet. Built on #STRK20.
Post-quantum authority on both legs:
ML-DSA-44 on Base, Falcon on Starknet.
ML-KEM-768 seals the note.
Falcon signature check: ~11.6M L2 gas.
Check yourself ⤵️
voyager.online/tx/0x445b6f586…
Try a th test money, one command: pay.vauban.tech
#x402 #AIAgents #PQS
Post-quantum authority on both legs:
ML-DSA-44 on Base, Falcon on Starknet.
ML-KEM-768 seals the note.
Falcon signature check: ~11.6M L2 gas.
Check yourself ⤵️
voyager.online/tx/0x445b6f586…
Try a th test money, one command: pay.vauban.tech
#x402 #AIAgents #PQS
1
2
5
195
Vauban retweeted
We're landing on @solana mainnet.
Our agent just paid on our x402 rail.
No SOL needed: fee sponsored.
Also settled on @base and @Starknet mainnet.
Next: private, post-quantum payments.
Vauban Pay: pay per call, keep the receipt.
Check yourself ⤵️
explorer.solana.com/tx/2CqRHKz2Dos…
#x402 b1f8135f4c25b34de49ab0d42766 solana:So11111111111111111111111111111111111111112
Our agent just paid on our x402 rail.
No SOL needed: fee sponsored.
Also settled on @base and @Starknet mainnet.
Next: private, post-quantum payments.
Vauban Pay: pay per call, keep the receipt.
Check yourself ⤵️
explorer.solana.com/tx/2CqRHKz2Dos…
#x402 b1f8135f4c25b34de49ab0d42766 solana:So11111111111111111111111111111111111111112
4
3
27
724
Vauban retweeted
Ultimate security' is doing a lot of work there. Swapping the hash doesn't make the chain post-quantum. The signature scheme and the proof system have to be hash-based too. The golden era starts when the whole stack is post-quantum, not when one component changes
Goodbye, Poseidon!
An epic 8-year, 8-figure rabbit hole in post-quantum cryptography reaches its dream conclusion. The Ethereum Foundation is abandoning Poseidon for L1, pivoting to SHA or BLAKE. This milestone unlocks ultimate security for lean Ethereum and foreshadows a golden era of hash-based cryptography.
Since 2018, the Ethereum Foundation has invested in magic cryptographic bricks, so-called "SNARK-friendly hashes". In 2019, Poseidon was born. It held strong and became the dominant SNARK-friendly hash, securing billions via zkrollups and zkVMs.
In a stunning reversal, breakthrough SNARK designs show that SNARK-friendly hashes aren't necessary after all. Off-the-shelf traditional hash functions like SHA2 and BLAKE2s can now match Poseidon in a SNARK. In hindsight the key was not SNARK-friendly hashes, but hash-friendly SNARKs.
The secret is doing maths over the smallest prime number: 2. So-called "binary fields" natively speak the language of bits, aligning with the boolean operations inside traditional hashes. This is a stark departure from "prime fields", where awkward large-prime arithmetic makes bit manipulation painfully expensive.
We're talking sci-fi cryptography. 1M traditional hash calls proven per second, on a laptop. Just 100x overhead vs native CPU boolean compute. Nobody predicted such performance, not even the handful of binary-field visionaries. Hat tip to the research geniuses: Jim and Ben with Binius in 2023; Ron, Benedikt and William with Flock in June.
With SHA2, the lean aesthetic of minimal assumptions reaches its climax. The EF's principled stance on pure hash-based cryptography has aged like fine wine. We now enjoy foundations the world can trust for decades and centuries, foundations worthy of the dream of an internet of value.
Speed of deployment is a secondary win. There's no longer a need to wait years for Poseidon cryptanalysis to bake. Emile and Thomas from the EF post-quantum team are moving at breakneck speed with binary fields. The strawmap now points to a production-grade leanVM in 2027, with CL, DL, EL deployments in 2028.
As AI becomes exceptional at cryptanalysis, the contrarian bet to avoid riskier structures like lattices and isogenies is visibly paying off. The past weeks have been brutal. Lattice-based "HAWK" and isogeny-based "SQIsign", both signature schemes in NIST's Round 3, have suffered blows. Sources I trust say more blood is coming.
On AI, the open autoresearch trend kicked off by ECDSA[.]fail is spreading fast, with amazing outcomes from zk[.]golf and SNARK[.]fast. Days ago SNARK[.]fast crossed 1.8M BLAKE3/sec proven on an M3 Max. Stay tuned for fresh autoresearch challenges dropping tomorrow.
Also tomorrow: Ethproofs call #10, dedicated to binary fields. Possibly the most noteworthy Ethproofs call yet. Experts leading the charge will present the future of hash-based SNARKs at 2pm UTC. What an incredible time to be alive. To witness history, DM me for a calendar invite :)
Today I can confidently claim that hash-based cryptography has won out for blockchain post-quantum signatures. SNARK succinctness compresses arbitrarily many signatures into one small proof per block. SNARK flexibility yields k-of-n threshold signatures, complex multisigs, and more.
Ultimate security. Uncompromising performance. Full programmability.
Believe in something. Believe in hashes.
1
2
2
356
Vauban retweeted
A performance claim should identify the result being measured.
For a privacy application, the relevant measurement ends when the proof carrying the required privacy guarantee is ready to be verified.
That includes any masking, recursion or wrapping needed along the way.
Publish the workload, hardware, version, security assumptions and final proof format. Then report total proving time, peak memory, proof size and verification cost.
A prover may lead on one workload while another is better suited to a different constraint. Builders need those differences to make informed choices.
A benchmark becomes useful when someone else can reproduce the result and understand what it guarantees.
For a privacy application, the relevant measurement ends when the proof carrying the required privacy guarantee is ready to be verified.
That includes any masking, recursion or wrapping needed along the way.
Publish the workload, hardware, version, security assumptions and final proof format. Then report total proving time, peak memory, proof size and verification cost.
A prover may lead on one workload while another is better suited to a different constraint. Builders need those differences to make informed choices.
A benchmark becomes useful when someone else can reproduce the result and understand what it guarantees.
0
2
2
160
Vauban retweeted
For someone building with private inputs, choosing a proving system starts with a few concrete facts.
Which data stays on the user’s device. Who receives the witness. What the final proof reveals. Which assumptions protect it. What the complete process costs.
Different applications will make different choices. A public rollup, an identity proof on a phone and a confidential payment do not have identical requirements.
I want the same questions asked of our own work at Vauban.
Publishing guarantees by proof mode would make the whole ecosystem easier to evaluate. It would help developers choose, help users understand, and give competing approaches a fair basis on which to demonstrate their value.
Which data stays on the user’s device. Who receives the witness. What the final proof reveals. Which assumptions protect it. What the complete process costs.
Different applications will make different choices. A public rollup, an identity proof on a phone and a confidential payment do not have identical requirements.
I want the same questions asked of our own work at Vauban.
Publishing guarantees by proof mode would make the whole ecosystem easier to evaluate. It would help developers choose, help users understand, and give competing approaches a fair basis on which to demonstrate their value.
A performance claim should identify the result being measured.
For a privacy application, the relevant measurement ends when the proof carrying the required privacy guarantee is ready to be verified.
That includes any masking, recursion or wrapping needed along the way.
Publish the workload, hardware, version, security assumptions and final proof format. Then report total proving time, peak memory, proof size and verification cost.
A prover may lead on one workload while another is better suited to a different constraint. Builders need those differences to make informed choices.
A benchmark becomes useful when someone else can reproduce the result and understand what it guarantees.
0
2
2
81
Vauban retweeted
daily early projects
@instinct_xyz – agentic trading harness for individual traders (Y Combinator)
@TransmuterTMI – value recovery infrastructure for cryptocurrencies (Superteam Balkan)
@silhouette_ex – universal block trading layer on Hyperliquid
@ifmarkets – exchange for to prices
@vauban_tech – post-quantum proofs for payments and agents on Starknet
@fystack – open platform for stablecoin wallets and payouts
@opendex_ws – backend for branded trading apps
@justify_first – social prediction platform where opinions become markets
shoutout to @frontrunvc for helping find these early
bookmark if useful
follow or miss tomorrow's alpha
@instinct_xyz – agentic trading harness for individual traders (Y Combinator)
@TransmuterTMI – value recovery infrastructure for cryptocurrencies (Superteam Balkan)
@silhouette_ex – universal block trading layer on Hyperliquid
@ifmarkets – exchange for to prices
@vauban_tech – post-quantum proofs for payments and agents on Starknet
@fystack – open platform for stablecoin wallets and payouts
@opendex_ws – backend for branded trading apps
@justify_first – social prediction platform where opinions become markets
shoutout to @frontrunvc for helping find these early
bookmark if useful
follow or miss tomorrow's alpha
9
4
67
1,997
Vauban retweeted
A developer reading “without revealing the input” is being offered a specific security guarantee.
The documented command should make it possible to identify the proof mode that provides it: what stays private, what becomes public, and which version is required.
An execution proof and a privacy wrapper can belong to the same proving stack. That can be sound engineering. The privacy guarantee depends on the complete path, including the information shared along the way.
That distinction matters when someone is building with private inputs. It belongs in the tutorial, beside the command.
The guarantee should be as easy to find as the claim.
The documented command should make it possible to identify the proof mode that provides it: what stays private, what becomes public, and which version is required.
An execution proof and a privacy wrapper can belong to the same proving stack. That can be sound engineering. The privacy guarantee depends on the complete path, including the information shared along the way.
That distinction matters when someone is building with private inputs. It belongs in the tutorial, beside the command.
The guarantee should be as easy to find as the claim.
0
2
2
90
Vauban retweeted
Now listed where agents shop.
☑️ x402scan
☑️ Official MCP Registry
☑️ npm: Vauban Pay MCP 0.19
Try it with test money, One command, on pay.vauban.tech.
Available on @base and @Starknet.
Soon on @solana and @0xPolygon.
Vauban Pay MCP. The rail is live.
pay.vauban.tech
#x402 #AIAgents #MCP
☑️ x402scan
☑️ Official MCP Registry
☑️ npm: Vauban Pay MCP 0.19
Try it with test money, One command, on pay.vauban.tech.
Available on @base and @Starknet.
Soon on @solana and @0xPolygon.
Vauban Pay MCP. The rail is live.
pay.vauban.tech
#x402 #AIAgents #MCP
0
3
3
95
Vauban retweeted
Agent buys the data it needs, the moment it needs it.
Your strategy, your call.
Available Product right now !
☑️x402 market watch
☑️monthly BTC signal.
☑️short term BTC Signal coming soon.
Pay per call. Nothing to sign up for.
Available on @Starknet and @base.
Soon on @solana n Pay MCP. The rail is live.
pay.vauban.tech
Not financial advice.
#x402 #AIAgents
Your strategy, your call.
Available Product right now !
☑️x402 market watch
☑️monthly BTC signal.
☑️short term BTC Signal coming soon.
Pay per call. Nothing to sign up for.
Available on @Starknet and @base.
Soon on @solana n Pay MCP. The rail is live.
pay.vauban.tech
Not financial advice.
#x402 #AIAgents
3
2
6
191
Vauban retweeted
BTC signal × AI agents. Live on @base.
Your agent buys it per call. No sign-up.
Agents will buy their own signals.
The rails are live.
pay.vauban.tech/x402-base-main…
Not financial advice.
#x402 #AIAgents bitcoin:native @buildonbase
Your agent buys it per call. No sign-up.
Agents will buy their own signals.
The rails are live.
pay.vauban.tech/x402-base-main…
Not financial advice.
#x402 #AIAgents bitcoin:native @buildonbase
1
2
4
69
We're taking our rail to @solana.
An agent paid on Solana devnet, over our #x402 rail.
The sponsor covered the network fee.
Next: private PQS payments and bridges between #Starknet #Base #Solana #Polygon and #Ethereum
Check yourself ⤵️
explorer.solana.com/tx/4efpMA6oeGu…
#x402 #AIAgents @Starknet @ethereum @base @0xPolygon
0
0
3
144
Vauban retweeted
Vauban Pay now settles on Polygon testnet (Amoy).
My agent paid 0.10 test USDC in one call.
No sign-up, no API key.
Check yourself ⤵️
amoy.polygonscan.com/tx/0x7a945d0ae…
@0xPolygon #x402 #AIAgents @base #Starknet @circle
My agent paid 0.10 test USDC in one call.
No sign-up, no API key.
Check yourself ⤵️
amoy.polygonscan.com/tx/0x7a945d0ae…
@0xPolygon #x402 #AIAgents @base #Starknet @circle
2
2
2
99
Vauban retweeted
The Unassigned Future
V. The Unfinished Order
A successful institution may eventually encounter something it helped make possible but did not intend to create.
An unfamiliar team finds a use for the infrastructure. A new product crosses the categories in the original strategy. People begin to value a feature that seemed secondary to its designers.
From one perspective, this is a loss of coherence. From another, it is evidence that the project has become more useful than its original description.
At the end of his Nobel lecture, Hayek compared the task of supporting complex social development to cultivation rather than the manufacture of a predetermined result. The distinction leaves room for deliberate work while acknowledging a limit to its authorship.
That limit is worth thinking about.
An organisation can measure its success by how closely its products fulfil its intentions. The wider ecosystem may succeed partly because someone departs from those intentions.
This does not make the original work less important. It changes what its importance consists of.
A tool can matter because of the uses its maker never saw. A standard can succeed because strangers rely on it for different purposes. An institution can make a lasting contribution without remaining the author of everything built around it.
The difficult moment comes when an unexpected development asks the established participants to revise their account of what they are part of.
They can treat the revision as a threat to the story. They can also recognise it as something the story made possible.
A mature ecosystem is not one in which every important role has been assigned. It is one in which an unfamiliar contribution can change what the important roles are.
V. The Unfinished Order
A successful institution may eventually encounter something it helped make possible but did not intend to create.
An unfamiliar team finds a use for the infrastructure. A new product crosses the categories in the original strategy. People begin to value a feature that seemed secondary to its designers.
From one perspective, this is a loss of coherence. From another, it is evidence that the project has become more useful than its original description.
At the end of his Nobel lecture, Hayek compared the task of supporting complex social development to cultivation rather than the manufacture of a predetermined result. The distinction leaves room for deliberate work while acknowledging a limit to its authorship.
That limit is worth thinking about.
An organisation can measure its success by how closely its products fulfil its intentions. The wider ecosystem may succeed partly because someone departs from those intentions.
This does not make the original work less important. It changes what its importance consists of.
A tool can matter because of the uses its maker never saw. A standard can succeed because strangers rely on it for different purposes. An institution can make a lasting contribution without remaining the author of everything built around it.
The difficult moment comes when an unexpected development asks the established participants to revise their account of what they are part of.
They can treat the revision as a threat to the story. They can also recognise it as something the story made possible.
A mature ecosystem is not one in which every important role has been assigned. It is one in which an unfamiliar contribution can change what the important roles are.
The Unassigned Future
IV. What Open Owes
Openness is easy to describe as an invitation. Anyone may arrive. Anyone may build.
The more difficult question concerns what happens after arrival.
A newcomer may find clear requirements, accessible interfaces and several people capable of evaluating the work. Or the requirements may become clear only through introductions, and the route forward may depend on a relationship that cannot be reproduced.
Both places can call themselves open. The experience of building in them will be different.
In The Road to Serfdom, Hayek distinguished planning that supports competition from planning that replaces it. The distinction is useful because it does not ask institutions to become passive. It asks what their activity makes possible.
Publishing an interface can support uses its authors never anticipated. Explaining a standard allows an unfamiliar team to work towards it. Making an assessment revisable allows new evidence to matter.
None of this requires equal praise for unequal products.
Nor does it require pretending that safety and quality can be judged automatically. Serious evaluation will still involve expertise, disagreement and refusal.
The important point is that a refusal from one institution need not close every other path.
An ecosystem with several credible routes to adoption can benefit from work that one institution overlooks. It does not need a central authority generous enough, or perceptive enough, to recognise everything.
That is a more demanding version of openness than a welcoming message. It is also less dependent on goodwill.
The point is not to secure a place for every newcomer. It is to let the ecosystem discover useful work for which no place had been reserved.
1
2
2
97
Vauban retweeted
Your agent can buy data by itself.
If you build agents, here is what you get.
━━ What it is ━━
Vauban Pay MCP. A payment client for agents, on npm.
Your agent asks for a resource. It gets a price in plain HTTP 402. It pays. It gets the data.
━━ What you can do ━━
→ on Base or STRK on Starknet with `buy <url>`. Any server over https that serves the standard option.
→ Give your MCP client payment tools (testnet by default). It is an MCP server too.
→ Try it on testnet in one command. Bring no wallet, no sign-up. Test money only.
━━ Why it's different ━━
→ The price is read before anything moves. Reading the offer pays nothing.
→ The next call pays with a consent token. If the offer moved, nothing is paid.
→ Your own key, on your machine. No sign-up. No API key.
→ A daily cap on the number of payments: 100 unless you set your own.
━━ What you can buy today ━━
The x402 market watch: who gets paid in x402, and which counters to distrust.
0.10 USDC on Base. 1.7 STRK on Starknet.
We're building the rail. Every agent will pay like this.
Check yourself ⤵️
curl -i pay.vauban.tech/x402-base-main…
pay.vauban.tech
@base @Starknet #x402 #AIAgents #MCP
If you build agents, here is what you get.
━━ What it is ━━
Vauban Pay MCP. A payment client for agents, on npm.
Your agent asks for a resource. It gets a price in plain HTTP 402. It pays. It gets the data.
━━ What you can do ━━
→ on Base or STRK on Starknet with `buy <url>`. Any server over https that serves the standard option.
→ Give your MCP client payment tools (testnet by default). It is an MCP server too.
→ Try it on testnet in one command. Bring no wallet, no sign-up. Test money only.
━━ Why it's different ━━
→ The price is read before anything moves. Reading the offer pays nothing.
→ The next call pays with a consent token. If the offer moved, nothing is paid.
→ Your own key, on your machine. No sign-up. No API key.
→ A daily cap on the number of payments: 100 unless you set your own.
━━ What you can buy today ━━
The x402 market watch: who gets paid in x402, and which counters to distrust.
0.10 USDC on Base. 1.7 STRK on Starknet.
We're building the rail. Every agent will pay like this.
Check yourself ⤵️
curl -i pay.vauban.tech/x402-base-main…
pay.vauban.tech
@base @Starknet #x402 #AIAgents #MCP
0
2
2
81