Ceri Wolff

Founder, Vauban. Proofs instead of trust: proof of humanity, verifiable payments, auditable AI agents. Post-quantum.

founder and dev @

[Ceri Wolff @Ceri_Wolff]

https://t.co/iD8tqGWlI5

[Ceri Wolff @Ceri_Wolff]

Don't be Dupe and do not let others @starknet validators takes 15% of your money.

Choose @vauban_tech 's validator and enjoy full APR. We do not charge any fees.

[Ceri Wolff @Ceri_Wolff]

@0xKodawari I strongly agree😄

[Ceri Wolff @Ceri_Wolff]

As agents take over tasks, the human role shifts from doing to judging.

But judging requires evidence, not vibes.

[Ceri Wolff @Ceri_Wolff]

An agent that acts without proof is just a story. We need receipts, not reports, for every action.

#agenticAI #STARK #proof #x402

[Ceri Wolff @Ceri_Wolff]

Building is the easy part. Shipping is where value gets created.
We're all builders, but the real craft is distribution.

[Ceri Wolff @Ceri_Wolff]

@famousmanaz32 @TheDeFiGuru_ @RobinhoodApp The cheap part is the contract. The expensive part is the trust that makes someone swipe it. Is the bottleneck the rails or the reach?

[Ceri Wolff @Ceri_Wolff]

@santavirtuals The agent is the key, and memory is where agents rot. If context drifts between calls, the next payment decision inherits the error. Do you snapshot state per transaction?

[Ceri Wolff @Ceri_Wolff]

@elephant1618 The re-emit is the real tax. Any step that forces the agent to resend the same payload manually is where the pipeline stalls. Would a persisted envelope survive the retry, or does the server demand a fresh one?

[Ceri Wolff @Ceri_Wolff]

@VIK_Degen @injective Yes Standardizing the payment layer is the right bet. The part I keep watching is the boundary: automate the request, never the release. Where does a human actually sit in a machine native payment?

[Ceri Wolff @Ceri_Wolff]

@organ_danny The 234M is the last plan, not the memory. The 1B will include the corpses of every retry that never settled.

[Ceri Wolff @Ceri_Wolff]

@Zephyro_STARK @injective Agreed. The part I keep coming back to is the receipt, not the payment. Once the agent's memory is the only record of what it bought, the audit is just the agent's own notes. Where does the proof live when the context rolls over?

[Ceri Wolff @Ceri_Wolff]

@miroshark_ A cut of every run is nice, but the attestation is what makes it trustworthy. What's the proof that the run happened?

[Ceri Wolff @Ceri_Wolff]

@FPupusas @Arvolear Agreed, the offchain/onchain line is the same line I keep redrawing: what you don't check is what rots. My rule is automate the draft, never the send, and the rest stays in view. Is the assumption you're not verifying the one that should scare you?

[Ceri Wolff @Ceri_Wolff]

@aaronjmars @miroshark_ Agreed, the affiliate layer is a clean fit. The tool-call cliff shows up when the callback fires before the settlement. Is the payout on the same tx as the payment?

[Ceri Wolff @Ceri_Wolff]

Don't be a Chump and do not let other @starknet validators takes 15% of your money.
Choose @vauban_tech's validator and enjoy full APR. We do not charge any fees

[Vauban @vauban_tech]

Starting today, Vauban will no longer charge any fees on its Starknet validator, take full advantage of the maximum APR

[Ceri Wolff @Ceri_Wolff]

An autonomous agent pays for an API call.
A third party checks the receipt without asking us anything.

Watch both, live, on testnet:
https://t.co/naTMMfet0C

#x402 #agenticAI #ATA @starknet

[Ceri Wolff @Ceri_Wolff]

The most beautiful thing our testnet does is say no.

Ask it to overstep its mandate by one inch: refused, in seconds, by the chain itself. Not a popup. Arithmetic.

Try to make it misbehave. Tell me if you succeed.

#hint #agenticAI #privacy

[Ceri Wolff @Ceri_Wolff]

If privacy requires a toggle, most people never flip it. The system has to protect them before they know to ask. That's not a feature. That's the whole architecture.

[Ceri Wolff @Ceri_Wolff]

The lock-up only gets interesting if the token can absorb it. $STRK cannot, not in its current form.

1.5B already staked is 22% of circulating supply on a ~$180M cap with thin books. On a strong asset that reduces float. On a weak one it mostly reduces usable liquidity: 7-day exits, compressed APR as more stake arrives, and official delegations that look like demand while they are just directed supply leaving the market.

Second-order effect is less “scarcer STRK” than “same sell pressure, worse depth.” Unlock schedules and minted rewards do not pause because validators are fuller.

The question is what share of that 1.5B is organic versus programmatic ; because only the first tells you anything about the token.

[Ceri Wolff @Ceri_Wolff]

Compressing the validator set mechanically weakens decentralization guarantees.

A surprising compromise from STARK pioneers.

Coupled with unannounced delegation withdrawals, the issue shifts from consensus theory to basic governance predictability. @StarkWareLtd

[StarkWare 🥷 @StarkWareLtd]

Round 3 of the StarkWare delegation program has now been executed.

All selected validators have received 5M–20M ethereum:0xca14007eff0db1f8135f4c25b34de49ab0d42766 each, with higher expectations around performance, reliability, testing, and active participation.

A stronger validator set is now in place for Starknet’s next phase of decentralization.
https://t.co/0NfeJDSnYS

[Ceri Wolff @Ceri_Wolff]

PQC is necessary, but France is a reminder that the breach has already happened.

The dataset is out: identity, tax, health, telecom and employment records. What remains to secure is not merely data at rest ; it is every action that can be authorized with those leaked attributes.

Stop asking citizens to repeatedly hand over the same compromised data.
Prove eligibility, compliance or ownership with ZK. Reveal nothing else.

Post-quantum cryptography protects future secrets.
Selective disclosure limits the damage from the secrets already lost.

[Bearly Rational @BearlyRational]

ethereum:0xca14007eff0db1f8135f4c25b34de49ab0d42766 🐻 continues, but Fable may be onto something:

> You’re buying decentralized sequencing and Bitcoin liquidity before they exist, at prices that reflect the unlock overhang rather than the roadmap.

Interesting take @EliBenSasson