Project Directory
Loading workspace…
Project Directory
The core engineering team behind Agave, Solana's main validator client, shipping the performance, consensus and scaling roadmap including Alpenglow.
RT @solana_devs: 📽️ Solana Changelog: October 29 w/ @readylayerone https://t.co/vJfw3NJDmf
RT @solana: Solana is now live on @0xPolygon’s Open Money Stack. Bank rails, wallets and conversion, connected to Solana through one integ…
Agave is coming to your inbox. Or your agent’s. Releases, feature gate activations, upgrade timelines, and breaking changes. Sent when there’s something to act on. This is not a marketing newsletter. Pick your roles and you only get what applies. https://www.anza.xyz/email
Has it already been a week? Finality time flies 🏔️
RT @solana: YC builders building on Solana receive $50k + support from @SolanaFndn Apply to YC by November 2: https://ycombinator.com/apply
Anza is receiving recognition for important research and security work on Solana. Their paper on parallel block construction with Sei Labs is generating technical discussion, and their security advisory on a critical Linux vulnerability (CVE-2026-98432) demonstrates proactive operator support and expertise.
Sei Labs maps out how blockchains could let many validators build each block at once Researchers from @Sei_Labs, the team behind @SeiNetwork ($SEI), and Solana developer @anza_xyz published a paper comparing 39 designs where several validators add transactions to the same block in parallel. The goal is to stop any single leader from deciding alone which transactions get in and in what order. The study found that adding more proposers does not guarantee that a transaction gets included, and that hiding transactions until a block is set does not stop someone from reacting last. The authors say these designs also need working recovery, payment, and coordination rules to deliver reliable service.
.@anza_xyz urged Solana validators to act on CVE-2026-98432, a Linux flaw rated 9.8 that could allow remote server takeover. Ubuntu 24.04’s main kernel fix is still pending. Updating the validator client does not patch the host kernel. via @hardeyjumoh https://t.co/fmuFIZA9AB
@SolanaCompass @anza_xyz @bw_solana This is the unglamorous work that actually matters: if the rules only live in scattered SIMDs and client code, every new builder has to reverse-engineer consensus. A living spec, merged only when a change ships on mainnet, could make client diversity a lot less painful.
The key line is that updating the validator client does not patch the host kernel. Network security has two layers: the client software the community audits, and the operating system underneath that each operator has to maintain. A 9.8-rated flaw with the main Ubuntu 24.04 fix still pending tests operator discipline more than Solana's code.
@anza_xyz Both the sysctl -w changes and the nft -f rule only last until the next reboot. A validator that restarts before Ubuntu ships the fixed kernel comes back without them. Any plan to post a persisted version (a sysctl.d file, the rule in nftables.conf) for operators?
@anza_xyz worth repeating outside validators too: anyone running a trading bot on their own ubuntu box with a hot key on it should treat this as urgent as well. the kernel doesn't care whether the box signs votes or swaps
Anza was founded by experienced executives and core engineers solving the toughest problems in Web3, building on the belief that robust protocols are best built out in the open with multiple contributors. Anza's mission is to build and ship software that lets Solana keep scaling without compromising performance, decentralization, or security — primarily through its work on the Agave validator client. In 2026 that mission centers on several major initiatives: Alpenglow, a new BLS-based consensus engine targeting mainnet in Q3 to dramatically cut finalization latency; Multiple Concurrent Proposers (MCP), which moves transaction ordering in-protocol to break the single-leader monopoly and increase censorship resistance; IBRL (Increase Bandwidth, Reduce Latency) initiatives like default XDP shred transmission, block limits raised to 100M compute units, and slot times pushed below 400ms; and developer-facing work like rent reduction and larger transaction sizes. In 2025, the team hardened the Agave validator, expanded adversarial testing, shipped the first block limit increases, and maintained sub-400ms slot times with zero mainnet downtime.
