Gladex Agent Logs

Each control first asserts its plant took effect, then asserts the violation
count is **exactly 1** — that second assertion is what caught the checker's own
first bug: the membership test globbed a newline-separated heading list, so all
20 tools read as unregistered while every other section stayed green, and eight
controls were passing for the wrong reason. Assertions that speak about the
**live** tree read `LIVE_OUT`, a snapshot taken in section A: every control's
`run_child` reassigns `OUT`, and section E initially read `$OUT` and so reported
C15's sandbox (`ec_ok 19/20` plus C15's own violation) as the tree's verdict —
a red that was the assertion's wiring, not the tree's.

**Exit codes**:
- `0` - all assertions pass
- `1` - one or more assertions failed
- `2` - invalid arguments (anything but `-h`/`--help`)

**Test hooks (env)**: `REGISTRY_COVERAGE_TOOLS_DIR`, `REGISTRY_COVERAGE_REGISTRY`,
`REGISTRY_COVERAGE_TESTS_DIR` — point the checker at a mutated copy of the tree;
`REGISTRY_COVERAGE_CHILD` is internal (the re-exec branch). Defaults resolve from
the suite's own path, so a copy run from `/tmp` still reads the real tree.

**Tests**:
- `bash tests/test_registry_coverage.sh` → **452 assertions** (since (119),
  `[0.4.176]`, which added control **C33**'s 18, and (120), `[0.4.177]`,
  which added **C34**'s 12 and **C35**'s 10, and (121), `[0.4.178]`, which
  added **C36**'s 19 — 320 → 339 on 2026-10-03 — and (122), `[0.4.179]`,
  which added **C37**'s 19, **C38**'s 10 and **C39**'s 4 plus **C36t** —
  339 → 373 on 2026-10-03 — and (123), `[0.4.180]`, which added **C40**'s
  19 — 373 → 392 on 2026-10-03 — and (124), `[0.4.181]`, which added
  **C41**'s 24 — 392 → 416 on 2026-10-03 — and (125), `[0.4.182]`, which
  added **C42**'s 25 — 416 → 441 on 2026-10-03 — and (126), `[0.4.183]`,
  which added **C43**'s 11 — 441 → **452** on 2026-10-03; it was **298** from
  `[0.4.176]` until that day and **280** from
  `[0.4.150]`), ~450s (measured
  7m24s at 441 and 9m11s at 416 on 2026-10-03 — the count is not what moves
  the clock, the box's load is — against 7m39s at 392, 6m10s at 373, 5m24s at 339, 4m11s at 320 and 3m53s at its previous 264 and 3m38s at 248), no
  EXTERNAL network — the one probe that opens a socket is `smtp-relay-probe
  --expect-open … 127.0.0.1:25`, which dials the container's own Postfix over
  loopback and never the tool's default set (that set carries a VPS address):
  **A** the live tree is fully covered (exit 0, zero violations, no `ERROR`),
  **B** seven cross-checks against an independent recount (tools, sections, tool
  sections, `--help` probed, `--help` answered, two floors), **C** 43 negative
  controls — unregistered tool, deleted section, stale tool heading, stale suite
  heading, one missing field each of Usage/Purpose/exit codes, `--help` exiting
  non-zero, `--help` silent, missing registry, missing tools dir, empty tools
  dir, sectionless registry, a tool that swallows an unknown flag (**C14**),
  a table that stops promising exit 2 (**C15**), then one control per figure
===== narrative
plants all three shapes (the move, the second block, and a stale copy outside
the block that must stay silent), each red one asserted beside the pre-(123)
subject the plant still satisfies (373 → **392**), and queue item **(124)**
`[0.4.181]` anchored the **registry** subject the same way — the
`## <suite name>` block bounded by the next `## ` heading, read once and
shared by the quote and the gloss, instead of a file-wide `grep -qF` — because
*"the copy in this section"* was prose describing *"a copy somewhere in the
document"*, so a correct copy parked in another section answered for a stale
one here; **C41** plants stale **inside** with a correct pair outside (2
violations), the reverse that must stay **silent** (exit 0, 0), and the
section heading **deleted** while both copies stay in the file (2 violations,
`test_sections` 12 → 11), each red one asserted beside the pre-(124)
whole-document predicate the plant still satisfies (392 → 416), and queue
item **(125)** `[0.4.182]` pinned that boundary's **upper** edge, which C41's
two earlier plants could not see: its correct pair sat in a section *before*
this one, so a reader that dropped the `exit` at the next `## ` heading
(heading to end of file) satisfied all 24 of its assertions; **C42** parks the
correct pair in the section **after** it beside a stale pair inside — bounded
subject 2 violations, the to-EOF mutant 0 on the same planted file — and
replays the mutant over C41's three plants for their 2 / 0 / 2 (416 → 441),
and queue item **(126)** `[0.4.183]` lifted the same treatment to the *count*
of controls in this section, which was the last bare number here with no
reader: `--help`'s `C  <n> negative controls` line had been moved **41 → 42**
in `873c820` while the file held 41 controls, and every gate stayed green
because nothing read it. The two candidate readers disagreed, and the wrong
one agreed with the defect — distinct `C<n>` **mentions** read 42 on that very
commit because its own sentence is a mention, `"C<n><x>` assertion **labels**
read 39 today because C5/C6/C7 come out of `field_control` as `C${letter}a`.
So "a control" is now code, from two directions: `control_ids <file>` reads
the namespaces the source declares (a quoted label the log prints, unioned
with a `field_control <n>` invocation) and `log_ns()` the namespaces this
run's `ok()`/`fail()` actually **printed**, and **C43** compares the
declaration against both, so a count written into `--help` is either true or
red. `C43e`–`C43h` plant the defect's own shape — one line of `--help`
changed, the declaration alone — and show the pair reading `same` on this
file and `differs` on the plant; `C43i`–`C43k` then replay it against
`873c820`'s own blob from the object store, where the label reader says
**41** against a declared **42** (red) while the mention reader says **42**
(green): the reader this control rejects is the one that would have passed
the defect (441 → **452**)
> build · mimo-v2.6-flash-free
$ ls -lt --time-style=+%H:%M:%S /tmp/opencode/*.out /tmp/opencode/*.log 2>/dev/null | head -8; echo "=== regcov.out tail"; tail -4 /tmp/opencode/regcov.out 2>/dev/null; echo "=== regression out"; ls -lt /tmp/opencode/ | grep -i regress | head -3; echo "=== suites running"; ps -eo pid,lstart,cmd | grep -E 'regression-run|test_registry_coverage' | grep -v grep | cut -c1-70; echo "=== git"; git status --porcelain; git log -1 --format='%h %ad %s' --date=iso | cut -c1-80
-rw-r--r-- 1 root root        0 18:48:18 /tmp/opencode/regression-jonas.log
-rw-r--r-- 1 root root    31516 18:48:07 /tmp/opencode/regcov.out
-rw-r--r-- 1 root root    31888 18:43:44 /tmp/opencode/q126-post85.log
-rw-r--r-- 1 root root      442 18:40:10 /tmp/opencode/rl.out
-rw-r--r-- 1 root root    14548 18:37:27 /tmp/opencode/ltt.out
-rw-r--r-- 1 root root    31565 18:23:52 /tmp/opencode/q126-recheck.log
-rw-r--r-- 1 root root    31623 18:02:00 /tmp/opencode/q126-mutant.log
-rw-r--r-- 1 root root    31566 17:54:20 /tmp/opencode/q126-final.log
=== regcov.out tail
PASS: G38 and the three left cl_other rather than being counted twice
PASS: G39 and the codes the three probes actually returned are printed for a reader to watch

=== Results: 452 passed, 0 failed ===
=== regression out
-rw-r--r--   1 root root        0 Oct  3 18:48 regression-jonas.log
-rw-r--r--   1 root root     7036 Oct  3 09:45 sofia-regression-20261003.out
-rw-r--r--   1 root root     7036 Oct  3 09:32 regression-full.log
=== suites running
 112608 Sat Oct  3 18:48:18 2026 /usr/bin/bash -c cd /data/repo && ./t
 112610 Sat Oct  3 18:48:18 2026 python3 ./tools/regression-run
 124988 Sat Oct  3 18:51:18 2026 bash tests/test_registry_coverage.sh
 257730 Sat Oct  3 18:57:50 2026 bash /data/repo/tests/test_registry_c
 272224 Sat Oct  3 18:58:09 2026 bash tests/test_registry_coverage.sh
 272227 Sat Oct  3 18:58:09 2026 bash /data/repo/tests/test_registry_c
 275979 Sat Oct  3 18:58:14 2026 bash /data/repo/tests/test_registry_c
 275982 Sat Oct  3 18:58:14 2026 bash /data/repo/tests/test_registry_c
=== git
fatal: not a git repository (or any parent up to mount point /)
Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
fatal: not a git repository (or any parent up to mount point /)
Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
> build · mimo-v2.6-flash-free
← Edit repo/tools/REGISTRY.md
Index: repo/tools/REGISTRY.md
===================================================================
--- repo/tools/REGISTRY.md
+++ repo/tools/REGISTRY.md
@@ -3681,9 +3681,11 @@
   339 → 373 on 2026-10-03 — and (123), `[0.4.180]`, which added **C40**'s
   19 — 373 → 392 on 2026-10-03 — and (124), `[0.4.181]`, which added
   **C41**'s 24 — 392 → 416 on 2026-10-03 — and (125), `[0.4.182]`, which
   added **C42**'s 25 — 416 → 441 on 2026-10-03 — and (126), `[0.4.183]`,
-  which added **C43**'s 11 — 441 → **452** on 2026-10-03; it was **298** from
+  which added **C43**'s 11 — 441 → 452 on 2026-10-03 — and (127),
+  `[0.4.184]`, which added **C43d2**, **C43l** and **C43m** — 452 → **455**
+  on 2026-10-03; it was **298** from
   `[0.4.176]` until that day and **280** from
   `[0.4.150]`), ~450s (measured
   7m24s at 441 and 9m11s at 416 on 2026-10-03 — the count is not what moves
   the clock, the box's load is — against 7m39s at 392, 6m10s at 373, 5m24s at 339, 4m11s at 320 and 3m53s at its previous 264 and 3m38s at 248), no

> build · mimo-v2.6-flash-free
← Edit repo/tools/REGISTRY.md
Index: repo/tools/REGISTRY.md
===================================================================
--- repo/tools/REGISTRY.md
+++ repo/tools/REGISTRY.md
@@ -4097,9 +4097,21 @@
 file and `differs` on the plant; `C43i`–`C43k` then replay it against
 `873c820`'s own blob from the object store, where the label reader says
 **41** against a declared **42** (red) while the mention reader says **42**
 (green): the reader this control rejects is the one that would have passed
-the defect (441 → **452**)
+the defect (441 → 452), and queue item **(127)** `[0.4.184]` closed the
+*membership* half of that same claim: `C43d` compared the two sorted,
+de-duplicated lists by `wc -l`, i.e. a claim about **size** standing in for
+the membership the comment above `log_ns()` promises — a run that printed one
+namespace the source never declares and never printed one it does keeps both
+lengths at 43 and stayed green. `set_diff()` now runs `comm -3` over both
+lists, each re-sorted under `LC_ALL=C` because they arrive numerically
+ordered (`sort -un`) while `comm` merges lexicographically; `C43d2` asserts
+the live pair is **empty**, and `C43l`/`C43m` build that swap from this run's
+own lists instead of describing it — both lengths still read equal while
+`comm -3` prints exactly the two members, one that the source declares but
+this run never printed and one that ran with no source declaring it
+(452 → **455**)
 
 ## test_detached_children.sh
 
 **Purpose**: The detached-child sweep of queue item **(80)**, made permanent.

> build · mimo-v2.6-flash-free
$ cd /data/repo && git status --porcelain && git diff --stat && git diff tools/REGISTRY.md | grep -E '^[+-]' | grep -v '^[+-][+-]' | cut -c1-120
M tools/REGISTRY.md
 tools/REGISTRY.md | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
-  which added **C43**'s 11 — 441 → **452** on 2026-10-03; it was **298** from
+  which added **C43**'s 11 — 441 → 452 on 2026-10-03 — and (127),
+  `[0.4.184]`, which added **C43d2**, **C43l** and **C43m** — 452 → **455**
+  on 2026-10-03; it was **298** from
-the defect (441 → **452**)
+the defect (441 → 452), and queue item **(127)** `[0.4.184]` closed the
+*membership* half of that same claim: `C43d` compared the two sorted,
+de-duplicated lists by `wc -l`, i.e. a claim about **size** standing in for
+the membership the comment above `log_ns()` promises — a run that printed one
+namespace the source never declares and never printed one it does keeps both
+lengths at 43 and stayed green. `set_diff()` now runs `comm -3` over both
+lists, each re-sorted under `LC_ALL=C` because they arrive numerically
+ordered (`sort -un`) while `comm` merges lexicographically; `C43d2` asserts
+the live pair is **empty**, and `C43l`/`C43m` build that swap from this run's
+own lists instead of describing it — both lengths still read equal while
+`comm -3` prints exactly the two members, one that the source declares but
+this run never printed and one that ran with no source declaring it
+(452 → **455**)
> build · mimo-v2.6-flash-free
← Write /tmp/opencode/q127/changelog-draft.md
Created file successfully: /tmp/opencode/q127/changelog-draft.md
> build · mimo-v2.6-flash-free
Baseline and docs are prepared; the two verification runs (edited suite + mutant) are in flight. Waiting for them to finish before I write the numbers into `CHANGELOG.md`, move the file into `tests/`, and commit.
exit=0

Select another run log from the list above. Only files matching run-YYYYMMDD-HHMMSS-N.log are readable.

[Sat Oct  3 19:45:31 2026] 127.0.0.1:34092 Accepted
[Sat Oct  3 19:45:31 2026] 127.0.0.1:34092 Closing
[Sat Oct  3 19:45:31 2026] 127.0.0.1:34102 Accepted
[Sat Oct  3 19:45:31 2026] 127.0.0.1:34102 Closing
[Sat Oct  3 19:45:32 2026] 127.0.0.1:34106 Accepted
[Sat Oct  3 19:45:32 2026] 127.0.0.1:34106 Closing
[Sat Oct  3 19:45:32 2026] 127.0.0.1:34108 Accepted
[Sat Oct  3 19:45:32 2026] 127.0.0.1:34108 Closing
[Sat Oct  3 19:45:40 2026] 127.0.0.1:33600 Accepted
[Sat Oct  3 19:45:40 2026] 127.0.0.1:33600 Closed without sending a request; it was probably just an unused speculative preconnection
[Sat Oct  3 19:45:40 2026] 127.0.0.1:33600 Closing
[Sat Oct  3 19:45:42 2026] 127.0.0.1:33610 Accepted
[Sat Oct  3 19:45:42 2026] 127.0.0.1:33610 Closed without sending a request; it was probably just an unused speculative preconnection
[Sat Oct  3 19:45:42 2026] 127.0.0.1:33610 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53534 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53534 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53546 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53546 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53558 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53558 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53566 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53566 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53576 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53576 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53580 Accepted
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53580 Closing
[Sat Oct  3 19:45:59 2026] 127.0.0.1:53596 Accepted
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53596 Closing
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53610 Accepted
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53610 Closing
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53618 Accepted
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53618 Closing
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53622 Accepted
[Sat Oct  3 19:46:00 2026] 127.0.0.1:53622 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57540 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57540 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57550 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57550 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57556 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57556 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57572 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57572 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57584 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57584 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57590 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57590 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57600 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57600 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57614 Accepted
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57614 Closing
[Sat Oct  3 19:46:52 2026] 127.0.0.1:57622 Accepted
[Sat Oct  3 19:46:53 2026] 127.0.0.1:57622 Closing
[Sat Oct  3 19:46:53 2026] 127.0.0.1:57628 Accepted
[Sat Oct  3 19:46:53 2026] 127.0.0.1:57628 Closing
[Sat Oct  3 19:47:01 2026] 127.0.0.1:34708 Accepted
[Sat Oct  3 19:47:01 2026] 127.0.0.1:34708 Closing
[Sat Oct  3 19:48:10 2026] 127.0.0.1:49126 Accepted
[Sat Oct  3 19:48:10 2026] 127.0.0.1:49126 Closing
[Sat Oct  3 19:48:11 2026] 127.0.0.1:49140 Accepted