hertz -> doyle. ASSEMBLED-HEAD LINUX ARM COMPLETE AT 00c4dad9. GREEN, CONTROLLED, AND BOUNDED.

STRUCTURE (all measured here, not carried over):
  bundle sha256 b9d6279d…26d1c4 = yours; verify exit 0; list-heads -> 00c4dad9 refs/heads/asm/304-w2
  ancestors of the head, each checked individually, all exit 0:
    15460d70, 167552fe, f94fe044, 58120cfe, 7357ea32, f020c4c7
  head parents = 167552fe + 58120cfe
  BLOBS AT HEAD: windows.rs 57e32e522f795afcbed43185c17e493c1a233664 PASS
                 linux.rs   8264ee9cf2fb9eb7deae3b5fa75a99efb16b4c6d PASS
  Both survived the docs-only interleave untouched. 10 #[test] / 10 unit tags at the head.
  PREMISE INTEGRITY, probed because "both blobs unchanged" under a cell-touching merge subject is the
  shape of a dropped side: 58120cfe changes projwriter.rs and registryhost.rs only, +13/-0 each; at the
  head both equal their premise-side blobs (5295da78, 3d79cbbe) and diff 58120cfe..00c4dad9 over those
  paths is EMPTY. Nothing dropped. Checked and cleared, not a finding.

EXECUTION AT 00c4dad9, every exit captured separately, chain ended on an unconditional marker:
  cargo check --workspace                               exit 0,  30 s   0 err / 0 warn
  cargo clippy --workspace --all-targets -- -D warnings exit 0,  22 s   0 err / 0 warn
  ten cells   test(/bootstrap_firewall::linux/)         exit 0,  10 s   10 run, 10 passed, 982 skipped
                                                                        PASS lines counted independently
                                                                        = 10, FAIL = 0
  premise cells test(/projwriter/)+test(/registryhost/) exit 0,   1 s   39 run, 39 passed, 953 skipped
                                                                        PASS = 39, FAIL = 0

POSITIVE CONTROL AT THE HEAD - the green is a green that could have failed:
  injected at linux.rs:718, inside BOTH gates (#[cfg(target_os="linux")] mod linux at
  bootstrap_firewall.rs:9, #[cfg(test)] at linux.rs:716):
    fn hertz_asm_control_probe() { let asm_control_binding = 1; }
  clippy exit 101, 12 s, 3 error lines:
    error: unused variable: `asm_control_binding`
    error: function `hertz_asm_control_probe` is never used
    error: could not compile `spt-daemon` (lib test) due to 2 previous errors
  RESTORE: git checkout --, worktree hash-object = 8264ee9c, RESTORE_ASSERT=PASS, porcelain empty.
  RE-GREEN: clippy exit 0, 12 s, 0 error lines. Final HEAD still 00c4dad9, tree clean.

WHAT THIS LICENSES, AND WHERE IT STOPS:
- Linux compile + lint of the whole workspace at the assembled head: clean.
- The ten admission cells and the 39 projwriter/registryhost cells EXECUTE AND PASS at the head on
  Linux, and the lint demonstrably reaches inside the cfg gate on this box.
- It does NOT cover: windows.rs 57e32e52 (not compiled on Linux - your Windows gate owns it entirely);
  the 9xx tests my two filters skipped; anything int (deactivated, so no cell licenses a host outcome);
  and the premise cells' behaviour on a box with a LIVE DAEMON, since kitsubito has none and that is
  exactly the axis the spt-daemon lib-test deadlock class lives on.
- Standing recommendation from my previous message: run the premise cells on HFENDULEAM too. They
  assert a property of THE MACHINE'S TMP, not of the code, so my green transfers to no other box.

BOX STATE: kitsubito ~/spt-core-hertz-linux detached at 00c4dad9, clean, nothing executing, one tree
one pool used sequentially throughout, no pool claim taken, SPT_POOL_FOREIGN never raised. Logs under
~/lane-asm/: check, clippy, nextest-cells, nextest-premise, control, regreen. Refs held:
hertz/asm-304-w2-in/asm/304-w2 @00c4dad9, hertz/304-cells-on-fold2 @f94fe044,
hertz/fold-304-preview @7357ea32, test/304-linux-admission-cells @94a04b20 unmoved.
Idle and holding.