𝑹𝒂𝒆𝒆🦉

𝐂𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐨𝐫 @primus_labs

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

THE RESEARCH BEHIND ZKTLS: TWO PAPERS WORTH READING

Most discussions about zkTLS focus on what the technology does. Fewer explore the research that makes it possible.

@primus_labs references two important papers that offer a deeper look into the cryptographic foundations behind its zkTLS implementation.

Here’s what they cover, explained in simple terms.

1. THE PROBLEM THE RESEARCH ADDRESSES

TLS secures communication between a user and a website. however the information received from a website is not automatically accompanied by a cryptographic proof that a third party can independently verify.

This creates a challenge when someone needs to prove what a website actually returned.

Sharing login credentials is not a reasonable solution, while screenshots can be edited or manipulated.

The research question is:

How can someone prove what a server said to a third party without modifying the server or unnecessarily exposing the underlying data?

This is the problem zkTLS research aims to address.

2. PAPER ONE: GARBLE-THEN-PROVE

Lightweight Authentication of Web Data via Garble-Then-Prove
USENIX Security 2024 | ePrint 2023/964

This is the zkTLS whitepaper referenced in Primus.

Earlier approaches relied on computationally expensive, general-purpose secure two-party computation.

The researchers introduced a lighter technique called Garble-Then-Prove, designed to achieve the required security goals with significantly lower communication and computational overhead.

Their reported results include:

→ Approximately 14× less communication.
→ Roughly an order of magnitude less computation compared with the previous state of the art.

These results were measured using authentication experiments involving Coinbase and Twitter APIs.

The paper also explains how authenticated payloads can be converted into commitments that make subsequent zero-knowledge proofs more efficient.

3. PAPER TWO: QUICKSILVER

QuickSilver: Efficient and Affordable Zero-Knowledge Proofs for Circuits and Polynomials over Any Field
ACM CCS 2021 | Best Paper Award Runner-up

The second paper explores the proof system known as QuickSilver.

Primus identifies QuickSilver as the interactive zero-knowledge proof system used in its zkTLS implementation.

In simple terms, an interactive proof involves communication between a prover and a verifier. this explains why an attestor participates in the verification process.

QuickSilver focuses on making zero-knowledge proofs more efficient and affordable across circuits and polynomials over different fields.

Together, these papers provide useful insight into the research behind efficient web-data authentication and zero-knowledge verification.

4. HOW TO READ CRYPTOGRAPHY PAPERS

You don't need to understand every mathematical expression to get value from a research paper.

Here are four practical steps:

1. Start with the abstract to understand the problem and the authors main contribution.
2. Identify the baseline used to compare the proposed approach.
3. Examine the threat model to understand which participants are trusted and which are not.
4. Review the evaluation setup, including the hardware, network conditions and assumptions behind the reported results.

One important distinction: the 14× communication reduction and order-of-magnitude computation improvement are research results against a specific previous system.

5. THE BIGGER TAKEAWAY

One thing I appreciate about studying infrastructure like Primus is the opportunity to look beyond product descriptions and examine the research behind the technology.

Cryptographic claims should be traceable, measurable and open to scrutiny.

Understanding the papers, checking the baselines and distinguishing research results from product performance are all part of evaluating infrastructure properly.

Because in cryptography, understanding how something works matters just as much as knowing what it promises to do.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

Good morning CT.

𝗔𝗡𝗔𝗧𝗢𝗠𝗬 𝗢𝗙 𝗔 𝗣𝗥𝗜𝗠𝗨𝗦 𝗔𝗧𝗧𝗘𝗦𝗧𝗔𝗧𝗜𝗢𝗡
How to read a cryptographic proof like a receipt.

𝗪𝗛𝗬 𝗟𝗢𝗢𝗞 𝗜𝗡𝗦𝗜𝗗𝗘 𝗔 𝗣𝗥𝗢𝗢𝗙?

Most conversations about zkTLS focus on one idea: proving a fact without exposing the underlying data.

But there is another important question worth asking:

What exactly does the proof contain, and how do we know what it actually verifies?

Understanding this matters because cryptographic proofs should not be treated as mysterious outputs that we simply accept.

A @primus_labs attestation is more like a digital receipt. It provides information about what was requested, which data was verified, who the proof was issued to and when the verification took place.

𝗧𝗛𝗘 𝗞𝗘𝗬 𝗙𝗜𝗘𝗟𝗗𝗦 𝗜𝗡 𝗔𝗡 𝗔𝗧𝗧𝗘𝗦𝗧𝗔𝗧𝗜𝗢𝗡

Here are the fields worth understanding:

→ attestor & taskId: Identify the network attestor responsible for issuing the attestation and the corresponding verification task.

→ recipient: Specifies the wallet address for which the attestation was generated.

→ request: Contains details of the verified request, including its URL, headers, method and body.

→ responseResolve: Identifies the specific response fields being checked, including their keys and data paths.

→ data: Contains the verified data items, represented as a string that must be parsed as JSON.

→ attConditions: Defines the conditions the response was required to satisfy.

→ timestamp: Records when the verification took place.

→ additionParams: Holds additional parameters supplied through the SDK. It is empty in the documented example.

𝗧𝗛𝗥𝗘𝗘 𝗤𝗨𝗘𝗦𝗧𝗜𝗢𝗡𝗦 𝗘𝗩𝗘𝗥𝗬 𝗗𝗘𝗩𝗘𝗟𝗢𝗣𝗘𝗥 𝗦𝗛𝗢𝗨𝗟𝗗 𝗔𝗦𝗞

Understanding the fields is only the beginning. Knowing how to evaluate them is what matters.

1. Is it the right source?

Check the request URL. A proof can be cryptographically valid while referring to an endpoint that does not establish the information your application needs.

2. Is it the right data?

Examine the response path. Which exact value was verified? A proof is only as relevant as the information it establishes.

3. Is it recent enough?

Check the timestamp. A result that was accurate yesterday may not be sufficient for a decision today. Applications must define their own freshness requirements.

You should also confirm that the recipient matches the expected wallet and that the verified data and conditions align with your application's requirements.

𝗪𝗛𝗔𝗧 𝗔 𝗩𝗔𝗟𝗜𝗗 𝗔𝗧𝗧𝗘𝗦𝗧𝗔𝗧𝗜𝗢𝗡 𝗗𝗢𝗘𝗦𝗡'𝗧 𝗚𝗨𝗔𝗥𝗔𝗡𝗧𝗘𝗘

A valid attestation establishes that specific information was verified against a particular request at a particular time.

However, it does not automatically establish that:

- The data represents the complete picture.
- The original source is always truthful.
- The information remains current after verification.

This is an important distinction in application security.

As Primus developer guidance highlights, validating a signature alone is not the same as fully validating an attestation. Applications must also inspect the content and timestamp.

𝗧𝗛𝗘 𝗕𝗜𝗚𝗚𝗘𝗥 𝗧𝗔𝗞𝗘𝗔𝗪𝗔𝗬

Zero-knowledge technology is not just about generating proofs. It is also about making those proofs useful, understandable and independently verifiable.

When developers understand what an attestation contains, they can make better decisions about the data their applications rely on.

Don't just verify the proof. Understand what it proves.

That's how cryptographic evidence moves from something that looks like magic to something an application can responsibly use.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@_rumaisah Is this worth your attention?

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Fatimabadam @uxuycom Best of luck

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@devameer0 @primus_labs Primus is tackling an important part of that challenge.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Rolexbasntyx @primus_labs Primus is tackling an important part of that challenge.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@AdamsuMus @primus_labs Primus is tackling an important part of that challenge.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Satoru_Gojo077 @primus_labs Primus can help users generate verifiable proof of information from supported websites without exposing their entire account history.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@capsxbt @primus_labs Primus can help users generate verifiable proof of information from supported websites without exposing their entire account history.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Dredd_the_001 Primus help users generate verifiable proof of information from supported websites without exposing their entire account history.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Lilllly69 @primus_labs Primus handles the evidence verifying source authenticity.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Fatimabadam @IPXExchange Interesting... excited to see how this evolves

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Fatimabadam Great tech.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Fatimabadam @IPXExchange This is worth checking out.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@VladaArt555 gPrimus

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@VladaArt555 Primus handles the evidence verifying source authenticity.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@VladaArt555 @primus_labs Prove what matters

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Jahto3 gPrimus my bro.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Jahto3 Primus handles the evidence verifying source authenticity.

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Web3ridyanulh @primus_labs gPrimus

[𝑹𝒂𝒆𝒆🦉 @Raeederth]

@Web3ridyanulh @primus_labs Prove what actually matters

[Bruno Jr Talent @bruno_jr_talent]

𝗧𝗵𝗲 𝗡𝗲𝘅𝘁 𝗙𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗢𝗳 𝗖𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝘁𝗶𝗮𝗹 𝗖𝗼𝗺𝗽𝘂𝘁𝗶𝗻𝗴

Imagine sending money onchain without revealing the balances involved.

The privacy sounds useful, but there is a technical problem behind it. The system still has to perform the necessary calculations while the information remains encrypted.

That means checking balances, updating values, and making sure the transfer follows the required rules without exposing the original data.

This is one of the biggest performance challenges in Fully Homomorphic Encryption (FHE).

And @primus_labs is working on making that process more practical.

𝗧𝗵𝗲 𝗛𝗮𝗿𝗱𝘄𝗮𝗿𝗲 𝗕𝗲𝗵𝗶𝗻𝗱 𝗣𝗿𝗶𝘃𝗮𝘁𝗲 𝗖𝗼𝗺𝗽𝘂𝘁𝗮𝘁𝗶𝗼𝗻

FHE requires intensive mathematical operations. Unlike ordinary computation, the system cannot simply decrypt the information, process it, and encrypt it again if it wants to preserve confidentiality throughout the operation.

This is why hardware acceleration matters.

GPUs provide a practical starting point because they can execute large numbers of operations in parallel. Rather than waiting for specialized processors, developers can use available GPU resources to optimize encrypted workloads.

Primus is taking this route with its AutoHoG engine.

𝗔 𝗡𝘂𝗺𝗯𝗲𝗿 𝗪𝗼𝗿𝘁𝗵 𝗣𝗮𝘆𝗶𝗻𝗴 𝗔𝘁𝘁𝗲𝗻𝘁𝗶𝗼𝗻 𝗧𝗼

Primus has reported processing an encrypted 64-bit balance transfer in approximately 90 milliseconds, with throughput exceeding 210 transfers per second.

The benchmark used eight NVIDIA RTX 5090 GPUs.

What makes this interesting is the type of work being measured. The computation includes balance checks, subtraction, addition, and selection while operating on encrypted values.

However, the 210+ figure should not be confused with blockchain TPS. It measures encrypted computation throughput, not the time required for wallet signing, network propagation, or onchain confirmation.

That distinction is important when judging how close the technology is to supporting complete financial applications.

𝗪𝗵𝗮𝘁 𝗖𝗼𝗺𝗲𝘀 𝗔𝗳𝘁𝗲𝗿 𝗚𝗣𝗨𝘀?

General-purpose hardware provides flexibility, but FHE could eventually benefit from processors designed specifically for its mathematical workload.

This is where FPGA and ASIC research becomes relevant.

A 2025 research paper on FPGA-based TFHE implementation reported bootstrapping performance around 2.9 times faster than its tested 16-core CPU and 1.1 times faster than its tested GPU.

Bootstrapping is particularly important because it helps manage noise that accumulates during encrypted computation.

There is an important limitation, though. The FPGA design was still incomplete, with key switching and compiler support not yet implemented. The results also came from a specific hardware comparison, not a universal benchmark.

Still, it shows why specialized hardware is attracting attention.

𝗧𝗵𝗲 𝗕𝗶𝗴𝗴𝗲𝗿 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗖𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲

I think the interesting part of Primus's progress is that scaling FHE isn't just about buying more powerful GPUs.

The cryptographic scheme, compiler, and execution engine must work together. An improvement in one layer can help remove bottlenecks elsewhere.

GPUs offer a practical route for current development, while FPGA and ASIC designs could provide more specialized execution as the technology matures.

But performance alone won't determine adoption. Cost, reliability, security, and integration with real applications will matter just as much.

For confidential finance to become practical, encrypted operations must be fast enough to use, affordable enough to deploy, and reliable enough to trust.