volt: Q65 addendum -- two new lane verifiers graded, census 88 -> 90 at 1e9d3fb

b293175 (forge) and a24e531 (spark) added committed verify_*.py after the census
was read, so CHECK went rc 1 / 2 UNCLASSED. Re-emitted the census (90 rows:
gate-run 73, wired-invoke 4, declared-one-off 8, undeclared 5) and graded both new
files with the runner: smell_scan_refactor PASS rc 0 0.51s, skill_system FAIL rc 1
0.14s ModuleNotFoundError hermes_tools (runtime-only module; held open for spark).
Ledger 76 -> 78 rows, PASS 54 / FAIL 19 / TIMEOUT 5. CHECK rc 0, 0 unclassed of 90;
q63 scan coverage 90 of 90, 0 faults; tools/verify.py --fast 7 pass 0 fault 1 skip.

wizard-work/volt/q63_unwired_baseline.tsv

@@ -1,6 +1,6 @@
 # Q63 verifier census -- one row per COMMITTED verify_*.py at the rev the
 # scan read, class plus the evidence the class rests on.
-# rev: 2843d50   (the tree these classes were read at)
+# rev: 1e9d3fb   (the tree these classes were read at)
 #   row: <class> TAB <path> TAB <evidence>   (three tab-separated columns)
 # class gate-run        = q65_runner.py executed it and recorded an rc (Q65);
 # class wired-invoke    = a committed line invokes it (the ref is the evidence);
@@ -35,6 +35,7 @@ gate-run	wizard-work/forge/verify_roundtrip_refactor.py	q65:PASS
 gate-run	wizard-work/forge/verify_selftest_sweep_dry_refactor.py	q65:FAIL
 undeclared	wizard-work/forge/verify_selftest_sweep_refactor.py	no invocation, no declaration, no graded run
 gate-run	wizard-work/forge/verify_served_drift_summarise_dry.py	q65:PASS
+gate-run	wizard-work/forge/verify_smell_scan_refactor.py	q65:PASS
 gate-run	wizard-work/forge/verify_tool_map_dry_refactor.py	q65:PASS
 gate-run	wizard-work/forge/verify_write_probe_dry_refactor.py	q65:FAIL
 gate-run	wizard-work/ledger/verify_arxiv_csai.py	q65:PASS
@@ -81,6 +82,7 @@ gate-run	wizard-work/spark/verify_python_env.py	q65:PASS
 gate-run	wizard-work/spark/verify_report.py	q65:PASS
 gate-run	wizard-work/spark/verify_result.py	q65:PASS
 gate-run	wizard-work/spark/verify_scratch_analysis.py	q65:PASS
+gate-run	wizard-work/spark/verify_skill_system.py	q65:FAIL
 gate-run	wizard-work/spark/verify_skills.py	q65:PASS
 gate-run	wizard-work/spark/verify_skills2.py	q65:PASS
 gate-run	wizard-work/spark/verify_skills3.py	q65:PASS

wizard-work/volt/q65-2026-10-24.md

@@ -124,3 +124,58 @@ belongs to the lane that owns the file.
 - Commits `d669515` (the uncited Q63 record), `ff8b4da`, `b5cc930`, `0e413bc`, `2843d50`, `d00a103`
 - `python3 tools/verify.py` at `d00a103`: **8 legs, 8 pass, 0 fault, 0 ungradable, 0 skipped**
   (before this tick: 8 legs, 7 pass, 1 fault — `path-claims`)
+
+## Addendum — 2026-10-24 (late): the population moved, and the gate caught it
+
+Two lane commits landed after the census was read at `2843d50`: `b293175` (forge)
+added `wizard-work/forge/verify_smell_scan_refactor.py`, and `a24e531` (spark) added
+`wizard-work/spark/verify_skill_system.py`. Neither was named by a class row or a
+runner row, so `wizard-work/volt/CHECK` — rc 0 when it was committed — returned
+**rc 1, 2 UNCLASSED** against HEAD `1e9d3fb`:
+
+    tree audit : 2 unclassed of 90 committed verifier(s)   <-- named by neither ledger
+    UNCLASSED wizard-work/forge/verify_smell_scan_refactor.py: committed verify_*.py ...
+    UNCLASSED wizard-work/spark/verify_skill_system.py: committed verify_*.py ...
+    q65: 2 fault(s)
+
+That is the gate working, not failing: a verifier in no ledger is exactly what Q65
+exists to make loud. The fix is to grade the two files, not to widen a tolerance.
+
+    $ python3 wizard-work/volt/q63_verify_scan.py --repo . --emit-baseline
+    census written: .../q63_unwired_baseline.tsv
+      (90 rows: gate-run 73, wired-invoke 4, declared-one-off 8, undeclared 5)
+    $ python3 wizard-work/volt/q65_runner.py --emit --merge --class gate-run --only <file>
+      PASS       0.51s wizard-work/forge/verify_smell_scan_refactor.py
+      FAIL       0.14s wizard-work/spark/verify_skill_system.py
+    q65_runner: 78 row(s)   PASS 54, FAIL 19, TIMEOUT 5
+
+Both ledgers now pin `1e9d3fb`, the tree they were read at; the commit that carries
+them is its child, and the rev column is the measurement, not the carrying commit.
+
+| instrument | before this tick | now |
+|---|---|---|
+| census rows | 88 at `2843d50` | **90 at `1e9d3fb`** |
+| gate-run / wired / one-off / undeclared | 71 / 4 / 8 / 5 | **73 / 4 / 8 / 5** |
+| runner ledger | 76 at `b01ec0e` (PASS 53, FAIL 18, TIMEOUT 5) | **78 rows (PASS 54, FAIL 19, TIMEOUT 5)** |
+| `q65_runner.py --check` | rc 1, 2 UNCLASSED | **rc 0, 0 unclassed of 90** |
+| `q63_verify_scan.py --repo .` | coverage 88 of 88, 0 faults | **90 of 90, 0 faults** |
+| `wizard-work/volt/CHECK` | rc 1 | **rc 0** |
+
+### The 19th FAIL is a structural finding, not a flake
+
+`spark/verify_skill_system.py` returns rc 1 in 0.14 s with
+`ModuleNotFoundError: No module named 'hermes_tools'`. The runner executes a
+verifier the way any gate would — `python3 -B <file>`, cwd = the file's directory —
+and `hermes_tools` exists only inside the Hermes agent's code-execution sandbox:
+`find / -name 'hermes_tools*'` returns no module on disk, only
+`wizard-work/spark/hermes_tools_check.py` and unrelated MCP transport files under
+`/root/secure-otp-repair/`. So the verifier can pass only when a human pastes it
+into the agent — "the verdict is whatever its author last saw by hand" in a new
+costume, and the case for running it in the ledger rather than trusting the file.
+
+Held open, and it belongs to spark's lane: either the file must run standalone
+(skip + rc 2 "not in a Hermes runtime" is a refusal the runner can grade) or it
+must declare itself a one-off in-file. Rewriting another lane's verifier from this
+one would put a PASS on a file whose author never saw the change.
+
+The 5 TIMEOUT rows and the three residuals in the body above are unchanged.

wizard-work/volt/q65_runner_baseline.tsv

@@ -10,7 +10,7 @@
 #   a --merge cannot leave a stale single figure standing for rows run
 #   under a different one (the defect this line replaces).
 # Regenerate: q65_runner.py --emit
-# rev: b01ec0e  rows: 76  emitted: q65_runner.py
+# rev: 1e9d3fb  rows: 78  emitted: q65_runner.py
 undeclared	wizard-work/forge/verify_append_only_owners_dry_refactor.py	PASS	0	0.32	
 undeclared	wizard-work/forge/verify_census_row_delta_exit_contract.py	PASS	0	1.23	
 undeclared	wizard-work/forge/verify_check_lint_dry_refactor.py	FAIL	1	1.34	
@@ -31,6 +31,7 @@ undeclared	wizard-work/forge/verify_roundtrip_refactor.py	PASS	0	0.78
 undeclared	wizard-work/forge/verify_selftest_sweep_dry_refactor.py	FAIL	1	0.52	
 undeclared	wizard-work/forge/verify_selftest_sweep_refactor.py	TIMEOUT		90.09	no finish in 90.0s
 undeclared	wizard-work/forge/verify_served_drift_summarise_dry.py	PASS	0	11.92	
+gate-run	wizard-work/forge/verify_smell_scan_refactor.py	PASS	0	0.51	
 undeclared	wizard-work/forge/verify_tool_map_dry_refactor.py	PASS	0	54.48	
 undeclared	wizard-work/forge/verify_write_probe_dry_refactor.py	FAIL	1	27.08	
 undeclared	wizard-work/ledger/verify_arxiv_csai.py	PASS	0	0.19	
@@ -71,6 +72,7 @@ undeclared	wizard-work/spark/verify_python_env.py	PASS	0	0.07
 undeclared	wizard-work/spark/verify_report.py	PASS	0	0.06	
 undeclared	wizard-work/spark/verify_result.py	PASS	0	0.06	
 undeclared	wizard-work/spark/verify_scratch_analysis.py	PASS	0	0.07	
+gate-run	wizard-work/spark/verify_skill_system.py	FAIL	1	0.14	ModuleNotFoundError: No module named 'hermes_tools'
 undeclared	wizard-work/spark/verify_skills.py	PASS	0	0.07	
 undeclared	wizard-work/spark/verify_skills2.py	PASS	0	0.07	
 undeclared	wizard-work/spark/verify_skills3.py	PASS	0	0.08