Verified Platforms
Quick Links
Where to Stay Secure
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
A proof of reserve does not tell you your stablecoin is safe. It tells you that one number matches another: the reserves an issuer reports, set against the tokens it has minted. That match is real.
It is also smaller than it looks.
On June 16, 2026, KRWQ, the largest Korean won stablecoin, turned on a live Chainlink proof-of-reserve feed. Issued by IQ and Frax on Solana, it replaced periodic attestations with real-time, tamper-proof verification that the off-chain won backing the token exists, published block by block instead of once a quarter. The badge turned green. The word that traveled with it was safe.
Safe is the part the badge cannot sign for.
This is a Kodex walkthrough with Eunha, who reads what a proof actually proves before she trusts the word printed beside it. The walk covers proof of reserve from the ground up: what it checks, what it leaves untouched, and why a verified reserve and a frozen redemption can share one screen.
Eunha pulls up the feed. The badge is green. She reads straight past it, down to the line that says what the green is actually claiming.
Underneath the badge, a proof of reserve answers a single question: does the reported reserve match the circulating supply, right now?
The mechanism is narrow on purpose. An oracle (Chainlink, for KRWQ) takes a reserve figure from the issuer's custodian, reads the token supply directly off the chain, and publishes whether the two line up. Mint more tokens, and the feed expects more won behind them. Let the won figure slip below supply, and the feed is meant to flag it the same block. Continuous also means you can watch the relationship drift live: if supply climbs while the reserve figure stalls, the gap shows the moment it opens, not a quarter later in a footnote. "Tamper-proof" describes the delivery: the number is signed and hard to alter in transit. It does not describe where the number came from.
This still beats what it replaces. How a stablecoin holds its peg depends on the reserves actually being there, and for years the only window onto them was a periodic letter from an accounting firm, often months stale before anyone read it. A live feed is harder to dress up for one reporting date, and harder to quietly unwind the day after the snapshot. Eunha grants it that. A continuous check is a real improvement over a slow one.
Then she points at the word "reserve" and asks which kind of proof this is.
Exchange proof of reserve, the version that spread after FTX, adds up customer balances and shows an exchange holds enough coins to cover them. Issuer proof of reserve, the KRWQ version, checks the collateral behind a token against that token's supply. Different object. Different trust model. The badge looks identical on both, which is exactly why the distinction gets lost.
In both cases, what it certifies is not the quality of the reserves, but their quantity.
The green hides a split.
"The reserves exist and match supply" is one statement. "The reserves are yours to redeem at par, today, while everyone else is also asking" is a different statement. A proof of reserve, at its best, delivers the first. It is silent on the second. Eunha keeps them in separate hands, because the badge quietly merges them.
Picture the morning a stablecoin wobbles. Price slips to 0.97. You open the issuer's page and the proof of reserve is still green, still matching supply block by block. You press redeem. The request sits there, pending. Both things are true at once: the reserves match the supply, and your dollar is not coming back this minute. The proof was never measuring the thing you needed at that moment.
BitGo, which runs reserve infrastructure for a living, states the limit flatly: a proof of reserve "may confirm that assets existed at a specific point in time, but it does not automatically prove solvency, liquidity under stress, governance quality, or the absence of off balance sheet obligations." Three gaps sit inside that one sentence.
Solvency. The feed reads the reserve pile. It does not read the issuer's debts. A company can hold a billion in reserves, owe a billion and a half somewhere the feed never looks, and still publish "fully backed." Backed by what it owns, undone by what it owes.
Liquidity under stress. The feed reports a value, not a guarantee that the value survives a stampede. Reserves parked in instruments that are fine on a calm Tuesday can be unsellable on the morning everyone redeems together. BitGo's phrasing: "settlement delays, collateral restrictions, or banking interruptions may limit how quickly reserves can actually be accessed when redemption demand rises." A peg is a promise about the worst day. The feed only ever reports the calm one.
Encumbrance. This is the quiet one. Reserves can exist, match supply, and already be promised to someone else: pledged as collateral, lent out, rehypothecated. "Reserves may exist while still being pledged, restricted, or tied to other financial obligations," BitGo notes. A borrowed dollar and a free dollar weigh exactly the same on the feed. Only one of them is truly behind your token. From the green number alone, you cannot tell which one you hold.
So the question "what backs a stablecoin" has two layers, and a proof of reserve reaches only the first. It confirms the reserves are present. It says nothing about whether they are unburdened, sellable, and larger than what the issuer owes.
The pegs that have broken did not break because the reserves were fiction from day one. They broke because the reserves were the wrong assets, or already spoken for, or impossible to sell at the one moment redemptions spiked. Each of those would have left a proof of reserve glowing green right up to the edge. The number matched. The money behind it stopped behaving like money when it counted.
A coin can pass its proof of reserve every block and still fail you on the one block you need out.
Now follow the number back to where it starts.
The oracle does not fly to Seoul and count won in a vault. It receives a reserve figure, usually from the issuer's own custodian, and faithfully publishes that figure on-chain. The cryptography around the delivery is real. What it secures is the transport of the number, not the truth of it.
Eunha traces the arrow back one step further than the badge wants you to look, and lands on a hand passing over a figure.
This is the old wall between an attestation and an audit, rebuilt in oracle form. An attestation confirms that figures match at a moment. An audit examines whether the controls, the accounts, and the processes behind those figures hold up. Reserve quality, and the gap between attestation and audit, is the same wall a holder hits whether the report arrives as a PDF or a data stream. Real-time delivery does not climb that wall.
It delivers the attestation faster.
An honest oracle reporting a number it was handed is still only as good as the hand.
Picture a custodian that overstates the balance by ten percent. The oracle does not catch it. It reads the figure, signs it, and posts a clean green proof on-chain. Now the bad number wears the costume of independent verification, which is worse than no badge at all, because it buys trust the reserves have not earned. The feed did its job perfectly. Its job was never to check the custodian.
It is also why the name on a stablecoin tells you less than it appears to. Shared issuance stacks now mint many differently branded dollars from one engine, and the brand on the token is not the entity holding the reserves. A proof of reserve on any of those tokens verifies the same arithmetic against the same kind of issuer-supplied feed. The logo changes. The question underneath does not.
No. It means one risk got smaller and the others stayed exactly where they were.
A live feed shrinks the "the reserves were never really there" risk, and the "they unwound it right after the snapshot" risk. Those are worth shrinking. It does nothing to the risks that actually break pegs: an issuer underwater on its liabilities, reserves that cannot be sold fast enough, collateral already pledged elsewhere.
Eunha draws it as one grid. What do you want to know about a stablecoin, and which of those can a proof of reserve answer?
| The question | Can a proof of reserve answer it? |
|---|---|
| Do the reserves exist and match supply right now? | Yes. This is exactly what it checks. |
| Are the reserves unencumbered, not pledged or lent? | No. A pledged dollar still counts on the feed. |
| Can reserves be sold fast enough to cover a redemption rush? | No. The feed shows a value, not liquidity under stress. |
| Does the issuer owe more than it holds? | No. The feed reads reserves, not the full liability sheet. |
| Is the reported number independently counted? | Only as far as the issuer's feed is honest. |
One yes. Four no. The badge is green for the one.
Keep this separate from the risk a yield product adds, because the two are easy to blur. When a stablecoin pays you to hold it, a yield layer stacks counterparty risk on top of the peg: you start trusting whoever generates the return, not only whoever holds the reserve. Proof of reserve and yield risk point at different parts of the same coin. A green reserve badge speaks for neither.
You do not need to audit a custodian to use a proof of reserve well. You need to know which question it answered, then go find the others yourself. Four questions cover the ground.
Is this exchange proof of reserve or issuer proof of reserve? One says an exchange holds the coins you deposited. The other says a token has collateral behind it. Reading one as the other is the easiest mistake to make, and the badge invites it.
Who supplies the feed? An independent verifier counting the assets is a stronger claim than the issuer's own custodian handing a figure to an oracle. Find out which one is behind the green.
Does it show liabilities, or only assets? Reserves alone cannot speak to solvency. If the proof reports what the issuer holds and never what it owes, half the picture is missing by design.
Are the reserves attested as segregated, unencumbered, and liquid? Existence is the easy part. Whether the assets are walled off, free of other claims, and sellable under pressure is the part that decides if your redemption is real.
A proof of reserve is a real instrument. It moved KRWQ from a quarterly letter to a live feed, and that is progress worth having. It also answers one narrow question and stays quiet on the ones that decide whether a peg holds when it is tested. Read the badge for what it signs, and carry the rest of the questions yourself. The Survival Framework is built on that exact habit: knowing which proofs you hold, and which you are still owed.
A green checkmark is where the question starts. Not where it ends.
A way of showing that the assets backing a token match the token's supply. For a stablecoin issuer, an oracle compares a reserve figure to on-chain supply, increasingly in real time rather than through a periodic report.
No. It verifies that reserves exist and match supply at a moment. It does not prove the issuer is solvent, that reserves are liquid under stress, or that they are free of other claims.
A proof of reserve, like an attestation, confirms figures match at a point in time. An audit examines the controls, accounting, and processes behind those figures. Proof of reserve is the narrower of the two.
Yes. If reserves are pledged elsewhere, cannot be sold fast enough, or the issuer owes more than it holds, a coin that passes its proof of reserve can still fail to redeem at par.
Exchange proof of reserve sums customer balances to show an exchange can cover withdrawals. Stablecoin (issuer) proof of reserve checks collateral against a token's supply. Same badge, different object and trust model.
