hertz #304 lane status — owed items, two of three delivered.

1. FROZEN SHA: c468e9df on test/304-remote-friction in .worktrees/hertz-304 (parents 5fd683e9 <- 0cea7d64 <- d89a655f). Source hashes retained at .spt/preserved/hertz-304-w1-c468e9df/source-hashes.json.

2. NONEMPTY INVENTORY (spt-daemon lib) — DELIVERED. cargo nextest list -p spt-daemon --lib --locked --build-jobs 2 --profile ci-windows -E <15 exact test() terms> --message-format json. exit 0, 82.2s, cold pool .worktrees/hertz-304/target (real dir, not a junction, 5.9 GiB). test-count 976 enumerated, matched exactly 15/15 requested: webserve::tests x14 + shellchan::tests::vocab_check_bounds_ops_and_args. Executing binary recorded: target/debug/deps/spt_daemon-8ceb80bef5261bf4.exe. Evidence .spt/preserved/hertz-304-w1-c468e9df/spt-daemon-lib-list-corrected.{command.json,exit,raw,stdout,observation.json}. The earlier exit-2 attempt (Cargo -j passed to nextest) is retained beside it as spt-daemon-lib-list.* with its empty stdout — that zero is a filter/argument failure, not an absence of tests.

3. STILL OWED: inventories for spt/bin (5 rc::tests) and spt-daemon/docs_server_e2e (1 case). Both need their own compile. Not started — see capacity/quiet note below.

ADMITTED SUBSET (exact, this lane, from selection.json):
- spt-daemon --lib: 14 webserve::tests:: + shellchan::tests::vocab_check_bounds_ops_and_args [INVENTORIED 15/15]
- spt --bins: rc::tests::{viewer_byte_detach_keeps_controller_and_child, viewer_event_detach_keeps_controller_and_child, detach_keybind_semantics, detach_prefix_spans_chunks, key_event_step_detach_sm} [inventory pending]
- spt-daemon --test docs_server_e2e: serves_published_surface_and_refuses_escapes_and_writes [inventory pending]
EXCLUDED and NOT admitted: #297 (no regression module authored yet), #300, webserve_e2e, webserve_cross_node_e2e, attach_link_push_e2e (real-daemon-network fixtures, per-fixture isolation unresolved), twohost, W3, xtask tests, Unix-only symlink cases, IR92.

CAPACITY, volume and time named as you asked: C: free 247,894,224,896 bytes (230.86 GiB) at 2026-09-11T08:39:20Z, GetDiskFreeSpaceExW on C:\. Clears the >=96 GiB bar with margin. My own admission sample before each producer still governs; floor 32 GiB and 64 GiB growth ceiling are enforced in-loop by the runner script, not by eye.

On the 07:00Z figure of 19,979,182,080 bytes: I cannot attribute it from retained evidence. It appears in no file under .spt/preserved (exact-digit grep, zero hits), and every C: free sample I did retain is far larger (191,765,405,696 at script-lane start, 191,755,583,488 at end, 247,894,224,896 now). Treat it as unsourced in the register rather than as a C: free sample; I will not defend a number I cannot point at a file for.

QUIET: not quiet right now. At 08:39:20Z the scoped builder census shows Runner.Worker.exe pid 32408 (birth 08:37:04Z) with cargo.exe pid 13648 (birth 08:37:27Z) — that is the b12d4619 post-merge run you flagged, building into the runner's own target, so it is CPU contention and not a risk to my pool. I am sequencing the two remaining compiles AFTER that window rather than racing it, and will re-run the census with a positive control immediately before each.

Meanwhile I am authoring the #297 bootstrap_firewall regression module (source only, no cargo). No GO needed for that and none assumed for anything beyond the subset above.