A few weeks ago, our autoresearch agent flagged an issue in SIMD-0376, a @Solana proposal that had been approved and was being implemented, but wasn’t active on mainnet yet. The SIMD changes how validators verify Ed25519 transaction signatures. The goal is good: adopt the ZIP-215 rules used by Zcash’s ed25519-zebra verifier, making it possible to batch signature checks and verify transactions faster. Our agent found a dangerous edge case. Under the proposed rules, a 64-byte all-zero signature would be accepted for any message when paired with the all-zero public key. On Solana, the all-zero key has special meaning. It is displayed as 11111111111111111111111111111111, the System Program ID, and is also commonly used as a sentinel value for “no authority,” “immutable,” or “uninitialized.” Programs often therefore set an authority to zero with the assumption that nobody can ever sign for it. If SIMD-0376 had been activated unchanged, that assumption would no longer hold. Metaplex gave us a concrete example. We found 433 mutable metadata accounts whose update authority was set to the zero key. The proposed verifier could have made those disabled authorities signable again. We reported the issue to Anza before activation. A fix has been prepared, and the SIMD is being amended to reject these weak keys while keeping the performance benefits. Full writeup: hackhack.ai/autoresearch/sim…

70,102