Optimism proposes Upgrade 19 and the Karst hardfork

What happened

OP Labs submitted the off-cycle Upgrade 19 proposal, bundling the Karst execution-layer hardfork with new predeploy upgrade machinery, OP Contracts Manager v2, a new primary fault-proof implementation, and portal dispatch changes.

What changed

If approved and executed, permissionless fault-proof chains would change their respected game type from CANNON to CANNON_KONA for withdrawal proving; L2CM and OPCMv2 would change upgrade execution; the portal would dispatch claims by game type; seven Fusaka EIPs and a lower BN256 pairing-input limit would become active; and op-geth and op-program support would end.

What did not change

The proposal did not establish mainnet approval or activation during June. It stated that fee take, governor/servicer role separation, ossified gas limits, and direct fee-margin controls would not change, and it anticipated no downtime.

Confirmed

• CANNON_KONA was proposed as the respected game type for permissionless fault-proof chains.
• The proposal would make kona-client on Cannon the primary fault-proof program used for withdrawals.
• L2CM would add deterministic predeploy upgrades through ProxyAdmin.
• OPCMv2 would replace five manager functions with unified deploy and upgrade calls.
• The mainnet activation was scheduled for July 8, 2026, pending governance approval, outside the permitted event-date boundary.

Still open

• The governance outcome and mainnet execution after the June boundary.
• Final audit results, production contract addresses, upgrade transactions, and deployed component versions.
• Operator readiness for migration from op-geth and op-program to op-reth and kona-client.
• Whether the new fault-proof and portal-dispatch paths operated normally after activation.

What it means for an advisor

• Review the Optimism bridge memo's withdrawal-proof, respected-game-type, and upgrade-execution assumptions before recognizing Upgrade 19 as operative.
• Verify governance approval, deployed contracts, upgrade transactions, audits, prestates, and operator migration evidence on-chain.
• Do not infer that deterministic tooling or completed testing makes the bridge safer or approved.