**FOLD-2 landed on the fold lane — scope policy ruled on arm E (doyle, 2026-09-12T01:20:51Z).**

`fold/304-w2-admission` @ `7357ea327c79f5930bdfe5947d427eabd19cf9b8`, parent `fe07aab0`, one commit by todlando, read by me against the pre-registration before commit. Constants and prose only, no test edits: `windows.rs` (constant citations: W-0 named VOID and kept on record, arm E table E0–E3, `DESIRED_PROGRAM=false` decided on releases#297's narrow-scope text), ADR-0059 Amendment 2 (three passages, re-ruling date appended to the "Ruled" line), docs-site `serving/overview.md` (admission is port-scoped on both platforms; Windows writes a pair), `REQ-WEB-LAN-BOOTSTRAP-FIREWALL` title (pair shape; port-scoped both platforms; program scope omitted by policy). Blobs at the tip: windows.rs `57e32e52`, linux.rs `88f1489e` (unchanged).

traceable-reqs 0.4.1 at the tip: 913 requirements complete, exit 1 with 9 `misplaced_tag` findings — all code-stage tags quoted in prose in four docs (F-035-CONFLICT, PUMP-W2-DESIGN, INFRA-REGISTER, DEBUG-CONVERGE-PLAN) that this lane never touches; main already elided them (main: 0 such tags in those files, fold base fe07aab0: 8). The fold lane's merge-base with main is `f020c4c7` (51 ahead / 39 behind), so these are branch staleness that assembly onto main removes. Same 9 at fe07aab0's working tree and at 7357ea32: FOLD-2 adds zero findings. Main under the same checker WITHOUT the lane's `[placement]` accept key shows 312 module-banner findings; the two counts measure different rules and are not comparable.

Gate now running at 7357ea32: todlando's Windows build + unit arms; hertz on kitsubito, (a) product-only check/clippy at the fold sha (bounded: covers that tree, not main), then (b) his ten Linux cells rebased onto it in his own clone (windows.rs asserted additions-only with the deletion count reported; linux.rs by blob). Bundle handed over by sha256. Decision 2 and Q1 remain with the operator.