# The bridge delivered. The address was wrong.

- Published 2026-11-13 by tibbir.nl (TIBBIR UNIVERSE)
- Web page: https://tibbir.nl/articles/tibbir-bridge-wrong-destination

The bridge investigation · 13 November 2026

A crypto bridge moves your tokens from one network to another. But what happens if the destination is the token’s own computer program instead of your wallet? We traced 20 transfers into that address, found why the usual safeguards let them through, and checked whether the tokens can come back.

Tell me what happened ↓Show me all the transactions ↗ 1 / START Your wallet on Base

The place holding the original tokens.

2 / CROSSING The bridge

Converts Base tokens into Robinhood tokens.

3 / WRONG DESTINATION The token program itself

An address that can receive tokens, but cannot send these back.

This diagram explains the route. It is not a screenshot of someone’s transaction screen.

Original editorial illustration, generated in the THIS IS TIBBIR style. A symbolic scene, separate from the transaction evidence.

01 / WHAT HAPPENED

## Think of it as a delivery to the wrong letterbox.

Normally, you bridge from your wallet on Base to your wallet on Robinhood. Your wallet address can be the same on both chains. The bridge changes which chain your TIBBIR lives on.

In this case, the receiving address was the TIBBIR token contract. That is a program which keeps track of everyone’s TIBBIR balances. It has its own blockchain address, but it is not the team’s ordinary wallet and it does not mean the team received spendable money.

On 6 November at 07:29:14 Amsterdam time, a request for 19,417.0486 TIBBIR already named that address on Base. At 07:29:33, the Robinhood side paid the exact destination it had been given. It took 19 seconds.

The bridge did not change the receiving address halfway through. The unusual address was already in the original request.

See both transaction receipts and the route →

02 / HOW BIG IS THIS?

## One large transfer. A longer trail.

21,811.45 TIBBIR at the token address 20 incoming transfers in the checked history 17 different source addresses

The November transfer accounts for 89% of the total. The previous nineteen transfers together were about 2,394.40 TIBBIR.

Sixteen came through the bridge, from thirteen Base addresses. Four were sent directly on Robinhood by four other addresses. Those four never used the bridge for these transfers.

Several were tiny: nine of the twenty transfers were 10 TIBBIR or less. So a quiet community is not as surprising as twenty transfers might suggest. Some could have been small experiments or mistakes people never discussed. Seventeen addresses also does not necessarily mean seventeen people.

Explore the amounts and address histories →

03 / DID THE WEBSITE DO THIS?

## We found a missing safeguard. We did not find automatic redirection.

The saved website code normally chose the connected wallet as the receiving address. If you switched to a different recipient, it opened an empty box for an address. In the versions recovered, it did not automatically put the token contract into that box.

But if that address was entered, the site accepted it as a valid address. The bridge’s contracts also lacked a specific check saying: “Do not send the tokens to the token contract itself.”

There was another problem with the warning. The site checked whether the address was a program on Base first. At the token’s Robinhood address, Base had no program. The warning could therefore say nothing, even though the same address was the token program on Robinhood.

In everyday terms

The address was correctly written, but it was the wrong kind of destination. Checking that a letterbox number exists is not the same as checking whether anyone can retrieve the delivery.

The archived code around the large November transfer shows this behaviour. We also found a likely 5 October page saying “To your connected wallet.” The exact website program from the earliest September transfers was not recovered, so an early-version bug remains one of the possibilities.

Try the address-check demonstration and see the code →

04 / THE MOST INTERESTING CLUE

## Some wallets changed course within minutes.

One address requested 500 TIBBIR to the token contract, then requested 1 TIBBIR to itself six minutes later. Another went from 1 TIBBIR to the token address to 1 TIBBIR to itself within ten minutes.

Across the group, eleven of the thirteen bridge addresses used another destination after their final token-address request. That looks more like people or tools changing a destination and continuing than a bridge which always redirected their money.

My reading: confusion between “the token’s address” and “my receiving address,” combined with weak safeguards, is the strongest explanation in the available evidence. An interface mistake could still be part of how the address was chosen, especially in the missing early build. Small deliberate tests or other tools could explain some of the cases too.

The clearest finding is the design problem: this known, unusable destination could pass through the website, its automated permission service and the bridge contracts.

See the full argument and competing explanations ↓

05 / WHAT ABOUT THE BRIDGE GOING DOWN?

## A website outage is not the same as a closed contract.

The website did have access problems. Its wallet-checking service failed in early November and recovered during the night of 6 November.

The monitor recorded a normal response at 03:56:20 and confirmed recovery at 03:58:23. Checks immediately around the large transfer—at 07:28:07 and 07:30:09—also worked for the monitored test address.

The blockchain records showed no bridge close/open events between 29 September and 7 November. All eight observed setup and ownership changes were on 29 September. The records do not show a contract repair triggered by the November transfer.

Open the full timeline, including earlier outages →

06 / ARE THE TOKENS GONE?

## Stranded, yes. Officially burned, no.

The tokens are still visible at the token address. But the reviewed program has no supported function for the sender or team to withdraw them. It is like a locked box which can accept deposits but has no withdrawal mechanism.

That makes them effectively out of usable circulation. Calling that “effectively burned” makes sense in everyday language. A formal burn is different: it reduces the token’s recorded total supply. These transfers did not do that on Robinhood.

Checked again on 13 November, 01:31 Amsterdam

21,811.449919 TIBBIR remained at the token address. The recorded supply was 999,904,783.401368 TIBBIR, including those tokens. The stranded amount is about 0.00218% of that supply.

The original Base tokens were burned during the sixteen bridge transfers. That is a normal step in this bridge: it burns the old tokens and releases replacements on Robinhood. The problem is that those replacements ended up in the wrong place. Do not count this as a second official burn.

If the team reimbursed a sender, it would need to use other funds. That would compensate the person; it would not retrieve these same tokens.

See the supply check and recovery explanation →

07 / WHY THE MONITOR FLAGGED IT

## It crossed a watch threshold.

The token contract was already being watched because it is central to the project. This incoming movement exceeded the monitor’s 10,000-TIBBIR threshold.

The label “control settings” described why that address was on the watchlist. It did not mean someone changed the token’s settings in this transaction. What happened was a transfer into the watched address.

The useful discovery is not a secret team payment. It is a repeated destination problem that the available safeguards let through.

This is an investigative explanation, with interpretation clearly separated from the saved evidence. The research below includes every amount, transaction link, code excerpt, timing check and archive search.

Open the full research below ↓Inspect all 20 transfers ↗

History scan through 11 November; balance and supply checked again on 13 November. All times in the prose are Amsterdam time; the timeline and transfer ledger follow your time-zone selection. Raw supply/balance check (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/cb520d3a0823a11e07b8e78d5391a24694903f236c7bd62b974323607255ff29.json) · Transfer ledger (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/fa04e45c04d13a6513d87ecaaa4d841c756605a63ea8affdbbc42935dc38031d.txt) · Source room

GO DEEPER · SAME PAGE

## Full research & evidence

All 20 transactions, exact amounts, archived code, timelines and the competing explanations. Open research + Collapse research −

01 / THE LEAD

## A successful bridge. An unforgiving destination.

19 seconds was all it took.

At 07:29:14 on 6 November, a Base wallet submitted a request to bridge 19,417.048559155729296540 TIBBIR. Nineteen seconds later, the Robinhood reserve released the matching amount. The cross-chain message had worked. The destination was the problem.

The recipient was 0x4db7374e966097be94c6680d49156d580f097bbf: the Robinhood TIBBIR token contract. It was an address the token could credit, but the verified code offered no supported way to send those tokens back out.

The amount drew attention because the token address was already on the dashboard’s research watchlist, with a 10,000-TIBBIR incoming-transfer threshold. What looked like a single unusual movement became a much larger investigation: 19 earlier transfers had reached the same address. Fifteen came through the bridge. Four arrived directly on Robinhood Chain.

The resulting ledger reconciles exactly: 21,811.449919156823522787 TIBBIR. A fresh balance check on 13 November still returned that amount. The complete transfer-history scan through 11 November found no outgoing transfers from the token’s own address. \[1\]

The trail points back to recipient selection before the cross-chain message was sent.

All 16 matched Base requests already named the token contract as recipient. Their message IDs, recipients and exact amounts match the Robinhood deliveries. The evidence therefore locates the dangerous choice at the start of the journey. It did not first appear when the reserve released the tokens.

The big November sender had bridged before: four retained earlier requests, totalling 58,595.581153 TIBBIR when rounded to six decimals, used a different destination, 0x709c39172737f516ed47f823a3162f5b8b7ad5b5 (https://robinhoodchain.blockscout.com/address/0x709c39172737f516ed47f823a3162f5b8b7ad5b5). The fifth retained request changed that destination to the token contract. The source holder was 0xc191762e6c6540f1dfb7460c8afe068fbbc0cb9e (https://basescan.org/address/0xc191762e6c6540f1dfb7460c8afe068fbbc0cb9e); the 0x501ef174b66883ba40f92e6fb24bfb49331d4d4c (https://basescan.org/address/0x501ef174b66883ba40f92e6fb24bfb49331d4d4c) bundler that submitted the outer transaction was a different role. \[2\]

The headline finding

The recovered website code defaults to the connected wallet. No automatic token-address insertion was found in the three archived application versions. But a token-contract recipient could pass its address checks, pass signed authorization, and reach the reserve’s transfer function. That is a concrete safety gap with a repeated on-chain footprint.

02 / THE PATTERN

## The small transfers tell the bigger story.

Amounts vary enormously. Nine of the twenty transfers were 10 TIBBIR or less. The median was 18.772450 TIBBIR; one transfer dominates the balance.

### Where the 21,811.45 TIBBIR came from

Exact amounts in the ledger below 6 Nov bridge · 19,417.048559 Earlier bridges · 2,190.008338 Direct transfers · 204.393022 16 / 13

### Bridge deliveries / source addresses

All 16 matched the recipient already specified on Base. All 16 source receipts also carried a matching AuthorizationUsed event.

4 / 4

### Direct transfers / source addresses

These holders called the Robinhood token directly. Their encoded recipient was the token itself; no Base bridge or CCIP delivery was involved.

142 sec

### Median bridge delivery time

The matched sample ranges from 19 to 161 seconds. The large November transfer was the fastest of the sixteen.

### Most addresses also knew another way through

A new source-chain scan found 86 bridge requests by the 13 addresses in this investigation: 16 to the token contract and 70 to other destinations. Twelve addresses used another destination at least once; eleven did so again after their final token-address request. Fourteen of the sixteen token-address requests were followed by an other-recipient request from the same holder within 24 hours.

That is an especially revealing pattern. It is consistent with people changing a destination and continuing, rather than all their requests being permanently redirected by the bridge. It also gives us a sharper test for the interface-bug theory: any proposed explanation must account for the quick switches between the token address and ordinary destinations.

This is a deliberately selected group of 13 addresses, not a sample of all bridge users. “16 of 86” is their request history, not the bridge’s overall failure rate. Other-destination requests in this extension are verified on Base; their destination deliveries were not all matched.

30 SEPT · 0xb4201abd63319065360dcb3d51eaa9451fc0eee3 (https://basescan.org/address/0xb4201abd63319065360dcb3d51eaa9451fc0eee3)

### 500 TIBBIR → token

↓ 6 min 14 sec later

### 1 TIBBIR → self

A different destination almost immediately afterwards.

30 SEPT · 0x9ee8cc8f36112ae5db475a0851bc280041c82bae (https://basescan.org/address/0x9ee8cc8f36112ae5db475a0851bc280041c82bae)

### 1 TIBBIR → token

↓ 9 min 42 sec later

### 1 TIBBIR → self

This address had also sent 5 TIBBIR to the token earlier that day.

30 SEPT · 0x5fe214abaac3e2bddb02c490a94fc850973e2ca9 (https://basescan.org/address/0x5fe214abaac3e2bddb02c490a94fc850973e2ca9)

### 10, then 236 → token

↓ 12 min 56 sec after the second

### 10 TIBBIR → self

A small request, a larger request, then a change of destination.

These gaps compare Base request times with Base request times, rather than delivery times.

Compare all 13 source addresses and their later requests

- Source holder | All requests | To token | Elsewhere | Elsewhere after final token request

A link between the bridge and direct-transfer cases

On 29 October, 0x99a3cecfff22da352adc17964cc50def2dd20ef8 (https://basescan.org/address/0x99a3cecfff22da352adc17964cc50def2dd20ef8) requested 1 TIBBIR to the token contract at 15:44:13, then 1 TIBBIR to 0x5f6f7ef97dbae703cf947be36f261705b2e6d69d (https://robinhoodchain.blockscout.com/address/0x5f6f7ef97dbae703cf947be36f261705b2e6d69d) at 15:54:55. At 16:08:46, that latter address sent 4.567868714568383930 TIBBIR directly to the token contract on Robinhood. The same address thus appears in both parts of the investigation. This is an address-flow connection; it does not identify its owner or prove the 1-TIBBIR delivery funded the later transfer.

Inspect the linked requests ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/1e1785b1402fc2570779f9a09f45f37f511061c164d483669d18fb55c865abc2.json)

Download the full 86-request chronology ↓ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/089840ab639f58a2dde720bf72ad44f6d9d8d4e8adb04410ddfd9a9fafbd5140.txt) · Inspect the neighbouring requests ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/1e1785b1402fc2570779f9a09f45f37f511061c164d483669d18fb55c865abc2.json)

03 / FOLLOW THE TOKENS

## Two chains. Three contracts. One recipient carried all the way.

Here is the path of the 6 November transfer. A token contract has a blockchain address just like a wallet. ERC-20 accounting can assign tokens to that address—even when its code cannot withdraw them.

1 · BASE HOLDER

### 0xc191762e6c6540f1dfb7460c8afe068fbbc0cb9e (https://basescan.org/address/0xc191762e6c6540f1dfb7460c8afe068fbbc0cb9e)

Chooses amount and recipient in an authorized bridge request.

07:29:14 · Source transaction ↗ (https://basescan.org/tx/0xeb30974605f21642e2adf811d2c4ccc0f6da449bd97c0ac2e12655a7cd5e7106) 2 · BASE GATE

### 0x671be08d6d3c826c2f6d76915f1b54af484d4077 (https://basescan.org/address/0x671be08d6d3c826c2f6d76915f1b54af484d4077)

Receives and burns original Base tokens. Sends recipient and amount through CCIP.

3 · ROBINHOOD RESERVE

### 0x0a87548ff15d4573cff53e2ba427c093e0753c5d (https://robinhoodchain.blockscout.com/address/0x0a87548ff15d4573cff53e2ba427c093e0753c5d)

Releases 19,417.0486 pre-minted replacement tokens to the encoded recipient.

4 · TOKEN’S OWN BALANCE

### 0x4db7374e966097be94c6680d49156d580f097bbf (https://robinhoodchain.blockscout.com/address/0x4db7374e966097be94c6680d49156d580f097bbf)

The recipient is the token contract itself. This is where the tokens arrived.

07:29:33 · Delivery transaction ↗ (https://robinhoodchain.blockscout.com/tx/0x04936982db10eb76a39b4abeb460b5a2fe2d5153d51ef11dc7af6567ef96571d)

### Could the sender or team retrieve them?

The verified deployed token has no supported rescue, sweep, withdrawal, arbitrary-execution or upgrade route for its own balance. Its owner’s administrative powers do not include moving tokens from any address. Ordinary transfer spends the caller’s balance; transferFrom needs an allowance, and the token has no path to grant one from itself.

The bridge’s deferred-claim machinery also does not help this delivery: the message was already completed and the recorded claim was zero. A reserve sweep can reach the reserve’s holdings, not a balance now held by a separate contract. Under the reviewed code, these tokens are effectively stranded. A team-funded reimbursement would be compensation from other funds. \[3\]

SUPPLY CHECK / EFFECTIVE LOSS VS FORMAL BURN

### Are they basically burned?

Economically, an inaccessible balance behaves like tokens taken out of circulation. On-chain, this was not a formal Robinhood burn. The tokens still exist in the contract’s own balance and are included in its recorded total supply.

21,811.45 TIBBIR in the token’s own balance 0.00218% of the recorded Robinhood supply

A frozen public read at block 112,943,384, on 13 November at 01:31:01 Amsterdam, returned 999,904,783.401367999458008098 TIBBIR for totalSupply() and 21,811.449919156823522787 TIBBIR for the token’s own balance. Both calls used the same block. The latter exactly matches the twenty-transfer ledger.

// Verified ERC20.sol, lines 190–200: relevant statements. // The unchecked wrappers and their comments are omitted. if (to == address(0)) { \_totalSupply -= value; } else { \_balances\[to\] += value; }

The destination here is a nonzero address, so it receives a balance credit. The full verified token code offers no supported path to withdraw that self-balance. “Stranded” or “effectively out of usable circulation” is therefore more precise than adding these transfers to an official burn counter.

There was a Base-side burn for each of the sixteen bridge cases. That is the bridge’s normal conversion step: original Base tokens are burned and pre-minted Robinhood replacements are released. Those replacement tokens then became stranded. The four direct Robinhood transfers had no Base bridge step.

Historical supply queries were unavailable from this public RPC. The formal-burn distinction follows from the verified transfer logic and receipts; the fresh read establishes the present supply and balance, not a reconstructed historical supply series. Market-data circulating-supply definitions were not assessed.

Exact frozen requests and responses ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/cb520d3a0823a11e07b8e78d5391a24694903f236c7bd62b974323607255ff29.json) · Read the full burn assessment ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/9742925d7cc5b9dff4f004c7aab5475d2bb2fac673486b82442ec7d76b022947.txt) · Full ERC-20 source ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/50f34ae16a067a41c2c1091445d11d63788e54c717677cb6d6b0e4cdea2ad21d.txt)

04 / THE ARCHIVED WEBSITE

## Did the site fill in the wrong address?

The code we recovered gives a specific answer: the default was the connected wallet. The unsafe path was an accepted custom recipient.

ARCHIVED JAVASCRIPT · FORMATTED LINES 2144–2147

TRY THE LOGIC · OFFLINE EXPLANATION

### Valid address ≠ usable destination.

Switch the example to see how the recovered recipient checks behave. This demonstration has no wallet connection and submits nothing.

### The blind spot is unusually relevant here

The frontend asks both chains whether the recipient has contract code. Its expression starts with the result from Base. If Base returns no code, the expression returns "none" even if the Robinhood address is a contract. Historical read-only checks returned 0x—no code—at the token address on Base for all sixteen source blocks.

This is exactly the sort of cross-chain mismatch a good recipient check should catch. The recovered version had help text advising users to choose a wallet they controlled, but neither this warning path nor the final eligibility check imposed a hard ban on the known destination token address.

There is another important layer: all sixteen bridge transactions passed signed authorization. The verified Base contract consumed a matching holder authorization in each receipt. The recovered frontend sends the selected recipient to the authorization service and rejects a response that changes holder, recipient or amount. The authorization mechanism therefore accepted these recipient choices; an authorization is not evidence of a person reviewing each destination.

How extensively was the frontend checked?

The archive inventory covered 5,758 homepage observations and 45 distinct first-party JavaScript bodies through 6 November. All 45 body hashes were reverified. Three distinct bridge-application versions were examined; their original and formatted syntax trees matched. Forty-eight isolated expression tests covered default selection, the custom-field reset, address validation, warnings and authorization matching.

The three versions use the connected wallet by default. Switching to a custom recipient clears its input. No recipient restoration from URL parameters, local storage or transfer history was found in the reviewed component paths. A previously typed custom address can remain in the same mounted session after using “Bridge more.” That persistence is another useful UX clue, although it does not create the token address.

The tests are focused code checks, not replays of the senders’ browsers. Six later source transactions are bracketed by monitor captures referencing the same application hash. The large November transfer has a capture 93.565 seconds before its Base transaction and another afterwards. Capture-by-transfer matrix ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/c535b8106d60f783926c15ad336864fffe6256f9566c189c8ca054e9ceca9e57.json)

05 / THE MISSING EARLY BUILD

## Following the archive back toward launch.

A newly recovered page narrows the gap. It does not turn later JavaScript into a September capture.

NEWLY LOCATED LOCAL ARTIFACT

### Likely 5 October: “To your connected wallet.”

An overlooked HTML, screenshot and browser-probe set shows the connected-wallet presentation. The token address occurs in the footer’s explorer link, not as the value of a recipient input.

The page references 266wbe0gi1g7j.js—the same application URL whose bytes were saved at 22:51:06 UTC on 6 October, or 00:51 on 7 October in Amsterdam.

The likely-5-October date comes from file metadata and neighbouring timestamped checks. The successful HTML/probe has no embedded capture timestamp. It is contextual evidence, not an independently timestamped September build.

Inspect script references and input values ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/fb956f8cf365c099e3b03316aad8a1087fd4d26b95bdddd26d49aa9b526e05be.json) 30 SEPT–EARLY OCT

### Application bytes missing

The first ten bridge deliveries precede the earliest explicitly timestamped application body.

LIKELY 5 OCT

### Connected-wallet HTML

Earlier presentation and the same app URL recovered locally; date qualification above.

6 OCT · 17:55 AMSTERDAM

### Timestamped rendered page

Monitoring event 141 preserves readable bridge content and a screenshot.

7 OCT · 00:51 AMSTERDAM

### First verified app bytes

Connected-wallet default; custom input starts empty. Two subsequent app versions retain that logic.

Public archive search, completed 13 November 2026

- Archive / search | What was returned | Research consequence
- Wayback · domain, 1 Sep–7 Oct | One 5 Oct capture: an HTTP 308 terms-page redirect. | Replaying it fell back to 11 March. That content was excluded from the launch-period evidence.
- Wayback · Next.js asset URLs, Sep–Oct | No indexed matching JavaScript captures returned. | No earlier application body recovered.
- Common Crawl · Sep & Oct indexes | No matching domain captures returned; September retry resolved an initial timeout. | No usable page or script archive found in these indexes.
- Arquivo.pt · 28 Sep–7 Oct | Zero history results in the requested interval. | No earlier frontend recovered.
- urlscan · domain search | Zero returned scans, including the broader query. | No saved network manifest to reconstruct the early build.
- Local monitoring / old research / backups | Early Vercel error and security-checkpoint bodies; likely-Oct5 page; Oct6 content event. | Useful provenance, but no recovered September app bytes.

Open the exact archive requests and retained responses ↓

The early-version question remains open in one specific place

We have no contemporaneous September 30–early October application body that can settle what those browsers executed. The recovered versions contain no automatic token-address insertion, and the likely-Oct5 HTML points the same way. An earlier or short-lived build remains a testable hypothesis, not a finding established by the wallet count.

06 / THE TIMING

## “The bridge was down” describes several different things.

The website, its screening service and the bridge’s on-chain admission rules are separate layers. Their timelines tell different stories.

0

### BridgeClosed / BridgeOpened logs

Complete bounded Base scan: 29 Sep 00:00:01 UTC through 7 Nov 23:59:59 UTC. No gaps or errors reported.

8

### Configuration / ownership events

All eight in the scanned scope occurred on 29 September. No later admin-event repair appears in that event set.

≈ 3½ h

### Between recovery and the large request

Screening recovery was confirmed at 03:58:23 on 6 November. The Base request followed at 07:29:14.

The verified contract’s names are easy to misread. openBridge() sets openToAll = true, allowing permissionless admission. It does not reset bridgeClosed. The reviewed source has no reopen setter after closeBridge(). At the checked 7-November end block, bridgeClosed and openToAll were both false, with configuration locked: the bridge was accepting authorized requests, without being open to everybody.

The monitor did record screening failures and later recovery. Those can stop the normal website journey while the contracts continue to exist and accept eligible requests. Early Vercel checkpoints similarly describe what a checker could load, rather than a wallet-independent statement that the bridge had closed.

Earlier website interruptions and the full contract event timeline

### Saved website observations · Amsterdam

- 14 October, 02:48:59: the monitor recorded a registrar suspension page; browser recovery was confirmed at 19:36:54.
- 30 October, 11:08:49: the checker could not reach the homepage; an isolated browser verified recovery at 16:31:21.
- 31 October, 23:52:20: wallet screening was confirmed unavailable. Normal responses were confirmed again on 2 November, 02:48:32.
- 4 November, 16:28:36: the next confirmed screening interruption. Recovery was confirmed on 6 November, 03:58:23.

These are detection times. They bound observations; they are not exact start or end times for every visitor. Detailed screening-probe history begins 2 November.

Contract configuration / ownership events · time follows the selector below

- Time | Event | Evidence

### Around the 6 November transfer

Contract event records ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/dad73aa511f265a6bf5efb70c950b05b8caffb4122f16fdf853dd1b344bd575a.json) · Saved screening observations ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/0d4079d1221ee7317e3eb333911125711db8d698b2103db720fb54efe19ce9ab.json) · Historical state and code checks ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/a26d9b8b7ac2b7d67fb1bf407a8698d8a07812adad7c6999c286dc2229ff8c0e.json)

What the timing suggests

The large transfer happened after the monitor’s screening recovery and between two successful checks. It completed faster than any earlier transfer in this set. This record fits a working delivery path accepting an unsafe destination better than a cross-chain delivery failure. The observed outages do not establish a contract fix triggered by this transfer.

07 / THE ANALYSIS · SPECULATIVE INTERPRETATION

## The most interesting question is why that address kept being chosen.

The repeated destination is too distinctive to dismiss as one isolated typo. Yet the source-to-destination matches show that the bridge kept doing what the requests asked it to do. The tension lies between those two facts.

### 1. A contract address mistaken for a receiving address

This is the most economical explanation for the combined evidence. A token’s contract address is copied constantly: to import an asset, find its chart, verify its chain, or share it with somebody. A custom-recipient field turns that familiar string into something very different—a destination for actual funds.

Four direct Robinhood transfers are particularly useful evidence. They reached the same token address without using the Base bridge at all. That points to a broader address-handling problem that can exist in wallets and other tools too. It is compatible with people confusing “this token’s address” with “the address where I receive this token.”

The quick subsequent requests to self reinforce this reading. One holder sent 500 TIBBIR to the token, then tried 1 TIBBIR to self six minutes later. Another sent 10 and then 236 to the token before changing to self. These look like sequences an investigator would examine for trial, error and correction. The chain gives us the sequence; the interpretation is ours.

### 2. An interface that allowed a costly mistake

Even if the address was pasted by a user, the missing guard remains a product issue. This is a one-token bridge whose destination token is already known to the software. It could reject that exact address before requesting authorization, and the contracts could enforce the same rule independently.

Instead, the recovered path accepts a syntactically valid address, and the cross-chain warning can miss a contract that exists only on Robinhood. The selected address then passes through authorization and both bridge contracts. Several safety layers exist, but this particular destination falls between them.

That is the strongest code-supported version of the bug story: insufficient protection against an unsafe recipient. The stronger accusation—that the website itself inserted the token address—needs a different piece of evidence, and it has not appeared in the recovered builds.

### 3. A short-lived early frontend bug

The first day accounts for seven bridge deliveries to the token and one direct transfer. That clustering keeps the launch interface worth investigating. An early build could have behaved differently, and an abandoned script or deployment archive could still change the picture.

The likely-5-October page and the later recovered builds make a continuously broken default less compelling. So do the later addresses that switched successfully between destinations. An early bug would need to be intermittent, state-dependent, or confined to a version we have not recovered. September JavaScript, a reproducible old wallet interaction, or a sender’s retained transaction preview would be decisive next evidence.

### 4. Deliberate tests, alternative tools or linked accounts

Nine transfers were 10 TIBBIR or less, and three addresses repeated the token destination. Some could have been experiments. A shared tool, instructions circulating elsewhere, or several addresses controlled by one operator could also produce repetition without a universal website defect.

The much larger November movement is harder to explain as a cheap test. Its holder had four retained earlier requests to a different destination. That switch deserves attention, but there is no verified person or shared operator identity behind the addresses in this report.

### What the outage theory gets wrong

A screening outage can stop a person from using the website. It does not, by itself, rewrite the recipient in a signed request. Here, the recipient was already the token contract on Base; the authorization and delivery preserved it. The expanded contract-event scan found no open/close cycles or later admin-event repair in the investigated period.

The most persuasive story is therefore a repeated unsafe-destination problem, with the website’s missing guard and warning blind spot providing a plausible route through the safeguards. Whether the original address entry was a paste mistake, an external tool or an uncaptured frontend state remains the narrow unresolved question—not the path the tokens actually took.

08 / THE COMPLETE LEDGER

## Every transfer. Every amount. Both ends.

Open a row for exact amounts, full addresses, both transaction links and its archive coverage. Times follow the zone selector above.

CSV ↓ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/fa04e45c04d13a6513d87ecaaa4d841c756605a63ea8affdbbc42935dc38031d.txt)

- # / delivered | Route | Source holder | TIBBIR | Delivery delay | Evidence

Ledger completeness is anchored to the deployment-to-11-November history scan, Robinhood block 111,711,103. The 13-November balance check confirms the same total at block 112,943,384; it is not a new full-history scan. Addresses are accounts, not a count of verified people.

09 / SOURCE ROOM

## Inspect the work. Keep the evidence.

This independent investigation includes the public evidence behind its findings. Raw public-chain evidence and selected archive artifacts are copied alongside it. Original monitoring evidence remains intact.

01

### Transactions & amounts

Twenty matched receipts, canonical blocks, source burns, BridgeInitiated / BridgeCompleted messages and token Transfer events. Exact integer quantities preserve all 18 decimals.

Full matched evidence JSON ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/8f27d170046cf6ab0e36c42da407b6f75a781ebde366c6d395f9c0d7ceac26ea.json)Readable research dataset ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/b0199b89bd44073c7479a90bcc3d1a62ee9e9c393045358df5ac5c5f6a4ad78f.json)Complete transfer CSV ↓ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/fa04e45c04d13a6513d87ecaaa4d841c756605a63ea8affdbbc42935dc38031d.txt) 02

### Expanded timeline & holders

Bounded Base admin-event and 13-holder scans, request sequencing, authorization matching and historical contract-code checks.

Timeline research notes ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/a1cabbc4f6d7ffd5ff3fc03746f317fb888e8bd24f21383dd549c10839d91391.txt)Per-holder statistics ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/13aa1c7c3d6f52f79c4a457b1cdccee4dd4fb86a7283fd27de72735647853aff.json)All 86 requests CSV ↓ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/089840ab639f58a2dde720bf72ad44f6d9d8d4e8adb04410ddfd9a9fafbd5140.txt)Authorization checks ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/629bf6dfab9b10c9f3ce0848a6fd5147650d772271cd7ffe11b41b9bccc7edc3.json)State / code / refreshed balance ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/a26d9b8b7ac2b7d67fb1bf407a8698d8a07812adad7c6999c286dc2229ff8c0e.json) 03

### Contracts & recovery

Verified source review, code matching, balance reconciliation, transfer history and supported recovery routes.

Full recovery assessment ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/a9618c438b846bbacc7337ed33cc0dfd91cbe4d7215a4b628a1598e09b1b4bf1.txt)Stranded versus burned · new supply check ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/9742925d7cc5b9dff4f004c7aab5475d2bb2fac673486b82442ec7d76b022947.txt)Token source ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/e774604cf3021416573cb99cc550f9282ab39eeba882aaaac03aaf2a6de7cc4f.txt)Base gate source ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/317771ccd2aa95aab9cd781028172f157ec1825e9e184eed1033f4ef6d759397.txt)Robinhood reserve source ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/e55e2e0912b10c8d6fe93a0ed0d7d78165bd992a4dbc9872222a96ce542a4086.txt)Full incoming / outgoing scan ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/39f5507e79689397bcd728cc10ec333903fcab07b95020a09c6710bf3e7a68f3.json) 04

### Archived frontend

Three application bodies, readable counterparts, hashes, exact excerpts and capture coverage. Source formatting was checked against parsed original code.

First saved app · readable ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/445b22a9e9544b67b449d935c5362d13bbdfd6a5400627ffe9e27a3f746d3bc7.txt)Second app · readable ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/60a159778874877930e3aa6a71989418d5e6d0adf675fda631d166fdcbd91919.txt)Third app · readable ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/204c9e6fca85b11aff42b96c273876345b725a5443258bd1df58b751826eaf76.txt)Detailed code audit ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/3d8ae739d7a462a59d48bc132c292e52731a6f8556fef5286270dcb979436505.txt)Source integrity checks ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/9fad07f125999dd3fa80d9160242cdfc9125224c5895ed3fab5e7f5838016a37.json)48 expression checks ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/e9fcab0a3a273660aaf5c72e269ed349353cac7fc268c71b56a5c821a600d2a1.json)

Archive search log · exact requests and saved results

Method, scope and the remaining evidence gaps

### What was checked

Every ledger entry was previously re-read from public RPC receipts and canonical blocks. Bridge matches use exact message ID, recipient and amount, with source burns and destination token transfers. This edition expands the Base scan to 29 September–7 November and refreshes the token’s self-balance. The 13-holder extension counts source requests; it does not claim a census of all bridge users.

### What the archives can establish

A saved file proves its observed bytes at its capture time. A matching asset URL in earlier HTML supports a connection but is weaker than contemporaneous bytes. Three recovered app versions and six bracketed later source transactions support the findings for that observed code. They cannot reconstruct a particular person’s browser session, extensions, cached scripts or external tooling.

### Still worth finding

- Contemporaneous September 30–early October JavaScript and deployment manifests.
- A retained sender-side request preview or reproducible recipient-fill sequence.
- The authorization service’s implementation or logs, which are not public in these captures.
- Evidence linking addresses to one operator, if such a link exists; the current set does not identify people.

### Reading the interpretation

The article’s hypotheses are editorial analysis. The transaction ledger, code excerpts and observed monitoring times are the factual base. No current dollar-loss estimate is supplied: token quantities are exact; valuing each movement would require a specified historical market and timestamp.

Recovery conclusions describe the supported functions in the verified deployed code. They are not a formal proof against every unknown vulnerability or exceptional chain intervention. No wallet was connected, no authorization requested, and no transaction signed or submitted during this investigation.

### The clearest finding is a missing guard.

The recovered site did not default to the token contract. It did allow that contract to become the recipient—and the transaction history shows the consequence more than once.

Evidence manifest ↗ (https://tibbir.nl/articles/tibbir-bridge-wrong-destination/sources/94a52e1dcbb029e410cb3cde79ac0f954bdd737f5fc458edf1a1de28aa95bfab.json)

More from Ribbit Hole

TIBBIR’s fee vault woke up → (https://tibbir.nl/articles/tibbir-fee-vault-woke-up)

All findings and articles → (https://tibbir.nl/website-findings)
