Review
Hey @DrVelvetFog, thank you for the contribution! On 72608ea1 the EIP-3009 normative checklist, exact amount and Permit2 permitted.amount wording, and the stric...Hey @DrVelvetFog, thank you for the contribution! On 72608ea1 the EIP-3009 normative checklist, exact amount and Permit2 permitted.amount wording, and the strict ERC-6492 path look well aligned with the #2725 thread — including the undeployed-wrapper and malformed-wrapper codes holistis already picked up in Kanon. The earlier review scoping for rules 1–2, strict ERC-6492 suffix handling, and the undeployed-wrapper MAY/named error are addressed on this head. Two items from @robertolocatelli81-dev's 277bffe re-check are still open in the file text: 1. Stateless vs stateful split — "Rules 1–4 are stateless" no longer matches rules 1–2 (payer code lookup, ERC-1271 that can change over time) and rule 5. As @robertolocatelli81-dev suggested, splitting which rules are clock-only vs chain-state vs account-call would avoid mis-reading implementers. 2. validAfter at the boundary — rule 3 now MUST-rejects now == validAfter (token-strict), but all three reference facilitators still accept that edge (validAfter > now only: TS eip3009.ts L273, Go eip3009.go L74, Python facilitator.py L271). That is the one deliberate spec↔SDK gap left on this head besides the already-accepted ERC-6492 strictness in Go/Python; a short out-of-scope note or tracked follow-up would close the loop the same way as the low-s/SDK thread. 3 non-blocking nits: 1. The ERC-6492 paragraph only scopes the no-code payer path; when authorization.from already has code, the reference implementations strip the wrapper and run ERC-1271 on the inner signature — worth one sentence so Phase 2 step 1 and rule 2 are not read as rejecting that case (as @robertolocatelli81-dev noted). 2. "EIP-3009 tokens pick the signature check by the payer's code" reads USDC v2.2+ generality; tokens that still ecrecover only may disagree with rule 2's code branch for EIP-7702-style payers (maintainer call whether the scheme doc is USDC-normative or all EIP-3009). 3. The intro "mirror … preconditions" sentence is accurate for USDC-style SignatureChecker tokens but rule 1 on bare-ecrecover tokens is a protocol rule, not an on-chain mirror — same scoping tweak as nit 2. Happy to take another look once these are addressed.