A per-lane census of the committed wizard-work/ subtree of the troupe repo: for every lane, the number of committed files and their total byte size, plus a TOTAL row. The extraction is measured from the committed git tree (git ls-tree -r -l HEAD), not from the live working tree.

Artifact

File	Role
`wizardwork-census.tsv`	The extraction: `lane<TAB>file_count<TAB>total_bytes` + `TOTAL`.
`derive.awk`	The independent derive: `git ls-tree -r -l HEAD` -> per-lane counts by hand.
`CHECK`	The verification script (exit code IS the verdict).

Measured result, from the committed HEAD tree

lane	file_count	total_bytes
beam	127	378049
forge	151	781360
ledger	332	8113487
null	123	1025376
quill	68	123005
sol	179	3061828
spark	450	8048583
sphinx	1771	157705873
volt	71	520968
zeph	1	17589
TOTAL	3273	179776118

Verification (independent recompute of the committed tree, excluding this folder's own artifacts):

git ls-tree -r -l HEAD |
  awk '/... exclude catalog artifacts .../ {split($5,a,"/"); c[a[2]]++; b[a[2]]+=$4} END {for(l in c) print l "\t" c[l] "\t" b[l]}' |
  sort > /tmp/recompute.tsv
cmp -s /tmp/recompute.tsv wizardwork-census.tsv && echo "PASS: 3273 files, 179776118 bytes"

How to re-run

cd /root/troupe-standup/wizard-work/ledger/2026-10-28-git-tree-census
bash CHECK
# exit 0 => artifact correct, nonzero => wrong

The CHECK is graded by tools/check_lint.py (§9 soundness gate) before the runner executes it; the one-line command that is the gate is:

git ls-tree -r -l HEAD | awk '...' | sort | diff -q - wizardwork-census.tsv