MANTRA Chain is producing blocks again. That is not the same thing as a finished security story. The project’s own status incident says mainnet (mantra-1) resumed around 05:30 UTC on 22 August 2026 after a coordinated restart on v8.4.0, with no rollback between the halt and the restart. The useful question for users and venues is narrower: did the halt actually contain the bug, and have the rails that move the MANTRA token on and off the chain reopened?
Our desk snapshot at 2026-08-23T23:56:20Z put Bitcoin at $77,536 (24h +1%) and Ether at $2,459.96 (24h +1.9%). That tape is calm. It does not tell you whether a Cosmos-EVM incident is closed. Status pages and client tags do.
What happened
MANTRA’s status log is the primary clock. The incident opens at 23:44 UTC on 20 August 2026 with a chain halt: endpoints and transactions frozen while the team reviewed root cause. Later updates said an attacker exploited a vulnerability in an upstream dependency used by the chain, and that the pause was precautionary so assets could not move while the network was down.
By 08:24 UTC on 21 August, the team said a patched release, v8.4.0, was in testing on DuKong, the public testnet. Validators were told to keep mainnet nodes offline until a coordinated restart. At 18:44 UTC that day, status said mainnet remained halted at block 17,449,398, a full state snapshot had been taken, store upgrades in the patch were empty (no module store changes, no state migrations), and the upgrade itself was expected to complete in seconds once the validator set moved together.
The same 18:44 UTC note is the most specific impact language on the page: based on analysis at that time, the incident affected two MANTRA-managed wallets, with no indication that user, exchange, or partner funds were affected. That is the team’s claim. It is not a third-party audit, and it is not a named CVE.
Restart: around 05:30 UTC / 13:30 HKT on 22 August 2026, mantra-1 was producing blocks again on v8.4.0. Status says the Cosmos-EVM module vulnerability was remediated in that tag, user balances were not altered, and there was no rollback. Public RPC including https://rpc.mantrachain.io was listed as operational in the 06:27 UTC monitoring update. DuKong testnet was still called out as offline, with a promise to bring it back over the following days.
The GitHub v8.4.0 release, published 16:51 UTC on 21 August 2026, matches that window. The changelog includes a bump of the mantra EVM dependency from v0.6.0-v8-mantra-3 to v0.6.0-v8-mantra-4, a v8.4.0 upgrade handler, an emergency upgrade height for DuKong, and adding the vesting module to the circuit breaker. It does not publish a vulnerability write-up. Binaries and a sha256sum.txt are attached. That is a client ship, not a post-mortem.
Venue residual, still live at our check: Kraken’s Mantra (MANTRA) funding incident remains in Investigating. The only public update is the 21 August 2026 06:55 UTC note that deposits and withdrawals may be delayed, with other funding methods operating normally. A green chain explorer and a red venue gate can coexist. They are different machines.
Context
A chain halt is an operational circuit breaker. It is not a proof that every related smart contract or module is healthy. MANTRA runs a Cosmos stack with an EVM module. When the team says “upstream dependency,” they are pointing at code they consume, not at a user fat-finger. That is why the GitHub bump of the mantra EVM crate matters more than any roadmap slide.
Empty store upgrades are a double signal. They mean the restart did not require rewriting account state, which matches the “no rollback” line. They also mean operators should not treat v8.4.0 as a balance-sheet event. If your MANTRA tokens were on-chain at halt height 17,449,398, the status page says that balance is still that balance. If your MANTRA tokens were in flight through an exchange deposit, you are in venue queue logic, not consensus logic.
That split is why counterparty risk sits next to protocol risk here. Kraken is explicit that the delay is the MANTRA funding gateway, not the whole venue. That is the honest framing: one asset’s on-ramp, not a site-wide outage. It is still enough to trap people who treated “chain is halted” as “send it to the exchange to be safe.”
Cross-chain wrappers and IBC-style bridges are the usual second blast radius after an L1 pause. MANTRA’s status list included bridge-migrate and IBC relay components among affected items during the outage window. The monitoring update says MANTRA-managed sites are mostly back and that explorers and indexers may lag. Lagging indexers are how people double-spend their attention: they see a stale nonce or a missing tx and send again.
Compare this with Ethereum client work we cover on the Ethereum history hub. Ethereum does not halt L1 because one execution client has a bug. MANTRA, like other smaller Cosmos app-chains, can stop the whole machine. That is a feature when you need to freeze an exploit. It is a cost when every dapp, every CEX deposit, and every IBC packet waits on the same validator coordination.
None of this is a Bitcoin range story. Use a market snapshot for BTC and ETH. Use the incident page for MANTRA rails. Mixing the two is how weekend headlines turn a module bug into a “crypto is down” myth.
Our read
Jonas here: show me the post-mortem, not the green status pill. I treat the 22 August restart as evidence that validators could run v8.4.0 together, not as evidence that the EVM-module bug class is gone from the stack, and not as evidence that every MANTRA deposit queue is healthy.
The team’s strongest operational claim is specific and checkable: no rollback, balances unchanged, halt height 17,449,398, restart ~05:30 UTC on 22 August on v8.4.0. I believe that until a later status note or explorer discrepancy contradicts it. The team’s weakest public claim is completeness of impact. “Two MANTRA-managed wallets” plus “no indication” about third parties is an interim sentence. Interim sentences age badly if they never get a technical appendix.
Falsifiable claim: by 31 August 2026, 23:59 UTC, MANTRA publishes a technical incident report that names the upstream Cosmos-EVM component and the patched commit or release line, and Kraken marks the Mantra funding incident resolved (or a later status note says deposits and withdrawals are operating). If that report says user balances were changed, or that a rollback did occur, this read is wrong. If Kraken is still Investigating on 1 September with no MANTRA write-up, the “restart equals closure” story some traders will tell themselves is the thing that failed, not the chain’s ability to produce blocks.
What to watch
First, the same status URL. Watch for a move from Monitoring to Resolved, a DuKong testnet restoration, and any walk-back of the two-wallet impact line. Second, the GitHub tag: if a v8.4.1 or a security advisory appears, that is more informative than another “we are monitoring” sentence. Third, venue pages. Kraken is the one we can cite now. If Binance, Coinbase, or others post MANTRA funding notes, treat those as separate gates, not as rumors on Telegram.
If you hold MANTRA tokens on-chain, the status page says you do not need to migrate. If you have a deposit or withdrawal stuck, the venue ticket is the ticket. Do not send replacement funds to a new memo because an indexer lagged. For how we triage incident pages versus headlines, the desk guide on following crypto news without getting played still applies. The blockchain news category is where we file this kind of client-and-halt tape.
A restart is a good outcome compared with a silent drain. It is not a substitute for the write-up that names the bug. Until that write-up exists, keep the incident open on your own checklist even if the explorer is moving again.