# "post-mortem" (exploit OR hack) — X 热门讨论 (2026-09-14 10:34 UTC)

## @TweetbyDominic (Lord Dominic) · 09-14 08:27 · ♥55 ↻0 💬64 Chainflip lost $736,442 in USDT through its TRON integration on September 12.

But the interesting part isn't the amount.

It's how the attacker made the protocol pay the same deposit twice.

The attacker found an issue involving TRON transaction memos.

Chainflip's validators had signed transactions, but manipulated memo data could cause the settlement system to interpret a transaction differently from what was originally approved.

The result:

One deposit could trigger multiple payouts.

The attacker tested the exploit, increased the amounts, and repeated it.

8 attempts were made over roughly 90 minutes.

6 produced unauthorized payouts.

Total extracted: $736,442.17 USDT.

Another $115,654.41 USDT swap was sitting in the vault but hadn't been paid out.

Chainflip halted trading and says affected users will be made whole.

But the bigger lesson isn't the $736k.

The attacker didn't break cryptography.

They didn't steal a validator's private key.

They exploited an assumption between two systems:

TRON's transaction model and Chainflip's settlement logic.

That's one of the biggest challenges in cross-chain infrastructure.

A validator can sign something correctly.

The blockchain can process it correctly.

The protocol can still interpret what happened incorrectly.

And that's enough to lose money.

Chainflip's architecture reduces some bridge risk by using native assets instead of wrapped tokens.

But the risk doesn't disappear.

It moves into the settlement layer.

Every new chain introduces a different transaction model, different metadata and different edge cases.

TRON had only recently become a Chainflip route, while USDT-TRC20 was added to Chainflip Lending on September 10.

Two days later, the TRON USDT route became the attack surface.

That doesn't prove the integration itself caused the exploit.

But it shows why adding a new chain is much more than connecting another network.

The attacker also didn't immediately try to drain everything.

They tested.

Observed.

Scaled.

Repeated.

That's how real exploits often work.

The goal isn't to find an impressive bug.

It's to prove the bug can become liquid money.

This is why the next Chainflip post-mortem matters more than the $736k headline.

Was this an isolated TRON memo issue?

Or does it expose a broader weakness in how Chainflip validates and settles cross-chain transactions?

That's the question.

Because $736k is measurable.

The potential loss of confidence isn't.

For users of cross-chain protocols, the takeaway is simple:

Don't judge security by TVL, volume or audits alone.

Ask:

Can one deposit trigger two payouts?

Which transaction fields are trusted?

Who authorizes the payout?

Can transaction metadata be changed without invalidating the settlement?

How does the protocol handle chain-specific edge cases?

Cross-chain security isn't only about securing the blockchain.

It's about making sure the application correctly understands what happened on the blockchain.

Chainflip just showed how expensive that distinction can be. https://x.com/TweetbyDominic/status/2099414613389394403