Gladex Agent Logs

+++ repo/tools/REGISTRY.md
@@ -4800,9 +4800,9 @@
 **Dependencies**: bash, python3 (stdlib), git; no network, no sqlite, no systemd.
 
 **Tests / mutations**: see `## regression-run` — 347 assertions, 14 external mutant runs (6 original + 7 baseline + 1 termination), each confined to its own section, plus the two **in-suite** plants (F's renamed heading, J's removed guard call, K's removed preflight call) and, since `[0.4.155]`, the two **docstring-shape** controls run through `REGRESSION_RUN_BIN` — the pre-fix replay (the sequencing paragraph unlabelled and unpaired, the tree as it stood before the step) → **327 / 4** with exactly `A17.7`, `A21`, `A22`, `A23` red, and an unlabelled column-0 block planted while every pair stays intact → **333 / 1** with `A23` the only red, which is A23's own value: nothing else in this suite reads a docstring block that has no label. A run whose `$TOOL` is not the shipped binary still loses exactly the **four** `G6*` assertions and nothing else — measured 2026-10-01 at **340 passed / 0 failed / 4 skipped** on `REGRESSION_RUN_BIN=<copy of the shipped tool>` (was **334 / 0**, the same four uncounted, at `[0.4.155]`, and **327 / 0** when the suite stood at 331), because G7–G11 are guarded by the shipped *location*, outside the shipped-*binary* condition — that run reads **343 passed / 0 failed / 4 skipped** since `[0.4.267]` (was **342 / 0 / 4** since `[0.4.158]`, **340 / 0 / 4** at `[0.4.157]`, **334 / 0** with the same four uncounted at `[0.4.155]`, and **327 / 0** when the suite stood at 331), because G13/G14 sit in the shipped-*location* branch too and run in both shapes.
 
-**Status**: Green ✅ 2026-10-01 (**346/346** with `[0.4.158]` — the (106) order-parity pair **A24**/**A25**, three measurements of that step: the shipped run **346 / 0 / 0 skipped**, the *pre-fix* tool replayed through `REGRESSION_RUN_BIN` → **341 / 1 / 4** with **`A24` the only red** (its `want`/`got` printing the two disagreeing orders) while `A25` stays green on both sides — `--help` is built from the tuple it is compared with, which is exactly why the claim needed two assertions — and a copy of the fixed binary → **342 / 0 / 4**. Was **344/344** with `[0.4.157]` — the (105) skip accounting, four measurements of that step: the shipped run **344 / 0 / 0 skipped**, the relocation the item describes **329 / 0 / 14 skipped** still exiting 0, that same relocation read *through the tool* **329 / 0 / 14** with `skipped_reported true` against **324 / 0 / `null`** pre-fix, a `REGRESSION_RUN_BIN` run **340 / 0 / 4**, and the symlink-missing control **342 / 1 / 1** with `G6c` its only red. Was **338/338** with `[0.4.155]`, the (103) eighth epilog pair (`sequencing:`) and its three needles **A21–A23**, of which A23 is a census of the docstring's column-0 blocks rather than a rendering check; three controls of this step, each measured: the **pre-fix replay** (the sequencing paragraph still unlabelled and unpaired, i.e. the tool exactly as it stood before the step, run through `REGRESSION_RUN_BIN`) → **327 passed / 4 failed**, reds `A17.7`, `A21`, `A22`, `A23` and nothing else; an **unlabelled column-0 block planted** while all eight pairs stay intact → **333 / 1** with `A23` the only red — the independent value, since A3/A4/A5/A16/A16b/A17 all stay green over a block they cannot see; and a plain **shipped-copy** run → **334 / 0**, i.e. still exactly the four `G6*` assertions short of the shipped run (338 − 4 = 334). Was **331/331** with `[0.4.154]` — the (102) shipped-location precondition and its isolation control; three measurements of the same step: the copy of `tests/` + `tools/` that went **322/4** before the guard goes **317/0** after it (`SKIP G: shipped-location checks (suite runs from /tmp/opencode/iso102b, not /data/repo)`, no `G1` line), a `REGRESSION_RUN_BIN`-shaped run still loses only the four `G6*` assertions at **327/0**, and the shipped run is 326/326 → **331/331** in 30.2 s → 60.1 s — was 326/326 with `[0.4.149]` — the (92) A16b pair-count equality; controls: an eighth pair spelled across two lines and placed FIRST in the tuple → **321/1**, the only red A16b (A16's floor still green, A17's pins still green, every derived loop still green — the blind spot reproduced with the new guard as its sole witness); the same two-line pair placed LAST → **320/2** (A16b plus A5.7, whose terminator set cannot know a label the extraction never saw) and the trailing comma dropped from the last pair → **316/3** (A16b, A17.7, A5.6); section G's shipped-location checks skipped in every control, since the mutant is not at `$REPO/tools/regression-run` — was 325/325 with `[0.4.148]` on 2026-10-01 (the (91) seventh epilog pair, A17.5 + A18–A20; control with that pair removed → **314/4**, the only reds A17.5, A18, A19, A20), and 318/318 with `[0.4.147]`, and 314/314 with `[0.4.132]` on 2026-09-29, and 307/307 with `[0.4.131]`, and 301/301 and 298/298 and 255/255 on 2026-09-29 before `[0.4.129]`'s section K, and 193/193 on 2026-09-26 — a figure that had already gone to 218 `[0.4.101]` and 237 `[0.4.102]` without this line moving)
+**Status**: Green ✅ 2026-10-08 (**347/347** with `[0.4.267]` — queue item (bj), the refusal's teeth: the shipped run **347 / 0 / 0 skipped** with `G0`'s census at **15** and `G13` green, the symlinked-source plant still refusing **rc 1 in 6 s** with no `$SB/iso`, and three out-of-tree plants of the extracted function each red exactly where it should be — `if false` → **346 / 1** with `G13 … (want=1:1 got=0:0)`, the cause line dropped → **346 / 1** with `G13 … (want=1:1 got=1:0)`, `if true` → no summary with `G14 … (rc=1 out=FAIL: isolation copy refused: …)`. Was **346/346** with `[0.4.158]` on 2026-10-01 — the (106) order-parity pair **A24**/**A25**, three measurements of that step: the shipped run **346 / 0 / 0 skipped**, the *pre-fix* tool replayed through `REGRESSION_RUN_BIN` → **341 / 1 / 4** with **`A24` the only red** (its `want`/`got` printing the two disagreeing orders) while `A25` stays green on both sides — `--help` is built from the tuple it is compared with, which is exactly why the claim needed two assertions — and a copy of the fixed binary → **342 / 0 / 4**. Was **344/344** with `[0.4.157]` — the (105) skip accounting, four measurements of that step: the shipped run **344 / 0 / 0 skipped**, the relocation the item describes **329 / 0 / 14 skipped** still exiting 0, that same relocation read *through the tool* **329 / 0 / 14** with `skipped_reported true` against **324 / 0 / `null`** pre-fix, a `REGRESSION_RUN_BIN` run **340 / 0 / 4**, and the symlink-missing control **342 / 1 / 1** with `G6c` its only red. Was **338/338** with `[0.4.155]`, the (103) eighth epilog pair (`sequencing:`) and its three needles **A21–A23**, of which A23 is a census of the docstring's column-0 blocks rather than a rendering check; three controls of this step, each measured: the **pre-fix replay** (the sequencing paragraph still unlabelled and unpaired, i.e. the tool exactly as it stood before the step, run through `REGRESSION_RUN_BIN`) → **327 passed / 4 failed**, reds `A17.7`, `A21`, `A22`, `A23` and nothing else; an **unlabelled column-0 block planted** while all eight pairs stay intact → **333 / 1** with `A23` the only red — the independent value, since A3/A4/A5/A16/A16b/A17 all stay green over a block they cannot see; and a plain **shipped-copy** run → **334 / 0**, i.e. still exactly the four `G6*` assertions short of the shipped run (338 − 4 = 334). Was **331/331** with `[0.4.154]` — the (102) shipped-location precondition and its isolation control; three measurements of the same step: the copy of `tests/` + `tools/` that went **322/4** before the guard goes **317/0** after it (`SKIP G: shipped-location checks (suite runs from /tmp/opencode/iso102b, not /data/repo)`, no `G1` line), a `REGRESSION_RUN_BIN`-shaped run still loses only the four `G6*` assertions at **327/0**, and the shipped run is 326/326 → **331/331** in 30.2 s → 60.1 s — was 326/326 with `[0.4.149]` — the (92) A16b pair-count equality; controls: an eighth pair spelled across two lines and placed FIRST in the tuple → **321/1**, the only red A16b (A16's floor still green, A17's pins still green, every derived loop still green — the blind spot reproduced with the new guard as its sole witness); the same two-line pair placed LAST → **320/2** (A16b plus A5.7, whose terminator set cannot know a label the extraction never saw) and the trailing comma dropped from the last pair → **316/3** (A16b, A17.7, A5.6); section G's shipped-location checks skipped in every control, since the mutant is not at `$REPO/tools/regression-run` — was 325/325 with `[0.4.148]` on 2026-10-01 (the (91) seventh epilog pair, A17.5 + A18–A20; control with that pair removed → **314/4**, the only reds A17.5, A18, A19, A20), and 318/318 with `[0.4.147]`, and 314/314 with `[0.4.132]` on 2026-09-29, and 307/307 with `[0.4.131]`, and 301/301 and 298/298 and 255/255 on 2026-09-29 before `[0.4.129]`'s section K, and 193/193 on 2026-09-26 — a figure that had already gone to 218 `[0.4.101]` and 237 `[0.4.102]` without this line moving)
 
 ## test_queue_source.sh
 
 **Purpose**: Hermetic suite for `tools/queue-source-check` — the gate that stops

$ cd /data/repo && grep -n "346" tools/REGISTRY.md | cut -c1-100; echo "=== 14 assertions refs ==="; grep -c "347" tools/REGISTRY.md
4767:- `bash tests/test_regression_run.sh` → **347 assertions**, hermetic (throwaway fixtures unde
4796:**Sections**: **A** `--help` contract incl. the **eight** docstring↔epilog byte-comparisons �
4804:**Status**: Green ✅ 2026-10-08 (**347/347** with `[0.4.267]` — queue item (bj), the refusal
=== 14 assertions refs ===
4
> build · mimo-v2.6-flash-free
$ cd /data/repo && grep -o ".\{70\}346.\{70\}" tools/REGISTRY.md
** behind the shipped-location branch, so 3 + 3 + 1 = 6); the **344 → 346** move is `[0.4.158]`'s section A (**A24**, the tuple's row order aga
's set comparison by construction could not see, so 1 + 1 = 2); the **346 → 347** move is `[0.4.267]`'s section G (**G13**, the planted-symlink
 and the "copy" would be the shipped tree — measured `[0.4.266]` as **346 / 0, rc 0 with `test -L $ISO/tests` true** before the refusal), and, 
tracted function each red exactly where it should be — `if false` → **346 / 1** with `G13 … (want=1:1 got=0:0)`, the cause line dropped → **346
ummary with `G14 … (rc=1 out=FAIL: isolation copy refused: …)`. Was **346/346** with `[0.4.158]` on 2026-10-01 — the (106) order-parity pair **
> build · mimo-v2.6-flash-free
$ cd /data/repo && grep -rn "346" tests/*.sh | cut -c1-120; echo "=== healthy run (post-reorder) ==="; bash tests/test_regression_run.sh > /tmp/opencode/run800-trr2.out 2>&1; echo "rc=$?"; grep -E '^=== Results:' /tmp/opencode/run800-trr2.out; grep -cE '^(PASS|FAIL): G1[34]' /tmp/opencode/run800-trr2.out
tests/test_regression_run.sh:610:# suite reported **346 / 0, rc 0** while `test -L "$ISO/tests"` was TRUE and
tests/test_regression_run.sh:626:# healthy total 346 -> 347 one assertion further down.
tests/test_regression_run.sh:639:# wide (census 14 -> 15, healthy 346 -> 347) and a tripwire adds no `PASS:`
=== healthy run (post-reorder) ===
rc=0
=== Results: 347 passed, 0 failed, 0 skipped ===
1
> build · mimo-v2.6-flash-free
$ set -u; cd /data/repo; d=/tmp/opencode/probe800bj/mutant; rm -rf "$d"; mkdir -p "$d"; cp tools/regression-run "$d/regression-run"; chmod +x "$d/regression-run"
REGRESSION_RUN_BIN="$d/regression-run" bash tests/test_regression_run.sh > /tmp/opencode/run800-mutant.out 2>&1; echo "rc=$?"; grep -E '^=== Results:' /tmp/opencode/run800-mutant.out; grep -E 'SKIP G6' /tmp/opencode/run800-mutant.out; grep -E '^(PASS|FAIL): G1[34]' /tmp/opencode/run800-mutant.out
rc=0
=== Results: 343 passed, 0 failed, 4 skipped ===
SKIP G6: shipped-location checks (under test: /tmp/opencode/probe800bj/mutant/regression-run)
PASS: G13 a planted symlink under $SB is refused and the cause is named (rc:cause-lines)
> build · mimo-v2.6-flash-free
$ cd /data/repo && (time bash tests/test_registry_coverage.sh) > /tmp/opencode/run800-regcov.out 2>&1; echo "rc=$?"; tail -3 /tmp/opencode/run800-regcov.out
Command moved to the background (shell ID: sh_11c3571a8001JJFHbg63q0JiyC).
Output is streaming to: /root/.local/share/opencode/shell/fa3e85b2145a60c127b5c9261e76a7ba6d520299/sh_11c3571a8001JJFHbg63q0JiyC.out
$ cd /data/repo && grep -n "^## \[0.4.266\]" CHANGELOG.md; sed -n '8881,8883p' CHANGELOG.md | cut -c1-200; echo "=== pointer text ==="; sed -n '8910,8911p' CHANGELOG.md
8881:## [0.4.266] - 2026-10-08 — queue item (bi): **the isolation copy in `tests/test_regression_run.sh` refuses a symlinked source** — with `$TESTS_DIR` a symlink the suite reported **346 / 0, rc 0 while `test -L "$ISO/tests"` was true and the "copy" was the shipped tree itself** (`-ef` true), i.e. G7–G11 were measuring the tree they claim to be a copy of; post-fix the same plant reads **rc 1 with one `FAIL:` line and no `iso` directory at all** (6 s), and through `tools/regression-run` it is `no_summary` → **exit 4**; healthy **346 / 0, rc 0** (count unmoved, `tools/REGISTRY.md`'s *346 assertions* still true), `tests/test_registry_coverage.sh` **464 / 0**
## [0.4.266] - 2026-10-08 — queue item (bi): **the isolation copy in `tests/test_regression_run.sh` refuses a symlinked source** — with `$TESTS_DIR` a symlink the suite reported **346 / 0, rc 0 wh

### Fixed
=== pointer text ===
### Queue
Authoritative next-candidate queue lives in `agent-logs/PROGRESS.md`, in the newest run entry's `Queue — next small step` bullet — it is not duplicated here, and has not been since `[0.4.67]`. To read the live queue, open `PROGRESS.md`; to know whether that is still true, run `tools/queue-source-check`.
> build · mimo-v2.6-flash-free
Writing the CHANGELOG entry `[0.4.267]`:
$ cd /data/repo && cat >> CHANGELOG.md <<'ENTRY'

## [0.4.267] - 2026-10-08 — queue item (bj): **the refusal gets in-suite teeth — `tests/test_regression_run.sh` extracts it into `refuse_symlinked_source()` and reads it twice** — `G13` plants a symlink under `$SB` and requires rc 1 *with* the cause on the line, `G14` (fail-only tripwire, deliberately uncounted) requires the same function silent on the source about to be copied, both run *before* the call site so a broken function cannot stop the run first; three out-of-tree plants each red exactly where they should be (`if false` → **346 / 1** with `G13 … (want=1:1 got=0:0)`, cause line dropped → **346 / 1** with `G13 … (want=1:1 got=1:0)`, `if true` → no summary with `G14 … (rc=1 …)`), healthy **347 / 0, rc 0** (census **14 → 15**, `G0`'s pin moved with it, `tools/REGISTRY.md`'s *346 assertions* → **347**), mutant-mode **343 / 0 / 4**, `tests/test_registry_coverage.sh` **464 / 0**

### Fixed
- **The gap `[0.4.266]` disclosed is closed: the refusal had no teeth in-tree.** Run 799's own honest limit read *"no in-suite assertion pins the refusal — a future edit could delete the guard and nothing in-tree would notice"*. The condition is now a function, `refuse_symlinked_source()` (`$1` = tests dir, `$2` = tools dir), a pure extraction: same `FAIL:` text byte for byte with the two paths substituted, same `exit 1`, no summary, no `iso` directory — the call site is `if ! refuse_symlinked_source "$TESTS_DIR" "$REPO/tools"; then exit 1; fi`.
- **Two readings, and the ORDER is the design decision.** `G13` plants `$SB/bj-symlinked-source` → `refuse_symlinked_source` and requires `rc:cause-lines` = `1:1`; `G14` calls the same function with the real source and requires `rc 0` with empty output. Both sit **before** the call site: after it, an always-refusing function would already have stopped the run, so the quiet half could never be observed — a dead assertion, which is the `(bh)`/`(bf)` vacuity class planted inside the very block that documents it. `G14` is additionally guarded by `[ ! -L "$TESTS_DIR" ] && [ ! -L "$REPO/tools" ]`: when the source *is* the symlink, silence would be the defect and the refusal below is the verdict (measured: on the symlinked-source plant `G14` stays out of the way and exactly one `FAIL:` refusal line prints).
- **Why `G14` is `fail`-only and `G13` is the one counted assertion — the queue's arithmetic, not a rounding.** G0's census counts `assert_* "G…"` / `ok "G…"` shapes inside this block: `G13` is one, `G14` is a tripwire that prints nothing when the guard is silent, so census moves **14 → 15** and the healthy total **346 → 347** — both numbers the queue item predicted. A second counted shape would have read **16 / 348** instead; the tripwire adds no `PASS:` line for the `Results:` line to account for, which is also why `G12`'s `PASS + FAIL + SKIPN = copy's passed + failed + skipped` still closes with the copy now reporting **15 skipped** rather than 14.

### Tests
- **The plant was re-run FIRST, on the pre-edit bytes, as the queue ordered** ("re-run the plant first (it costs 6 s) so the teeth are proven against the same harness, not a fresh claim"): `/tmp/opencode/probe799bi/` rebuilt from the tree as it stood before this edit → **rc 1 in 6 s**, refusal line, `MEASURE: no $SB/iso directory exists`. Rebuilt from the edited bytes → the same refusal, now preceded by `PASS: G13 …` (143 `PASS:` lines before the stop, was 142 — the +1 is `G13` itself), `G14` silent, still **no `iso`** and no `Results:` line (so `tools/regression-run` still reports `no_summary` → **exit 4**).
- **Three plants of the extracted function, out of tree under `/tmp/opencode/probe800bj/` (never committed), one verdict each:**

  | # | mutation | result |
  |---|---|---|
  | 1 | `if false` — never refuses | **346 / 1, rc 1**, `FAIL: G13 … (want=1:1 got=0:0)` |
  | 2 | cause line dropped, rc kept | **346 / 1, rc 1**, `FAIL: G13 … (want=1:1 got=1:0)` |
  | 3 | `if true` — always refuses | no `Results:` line, rc 1, `FAIL: G14 the refusal stays silent on a real source (rc=1 out=FAIL: isolation copy refused: …)` **then** the call site's refusal — both halves observably red |

  Each mutation was applied to a copy of the suite, run through the run-799 harness's four disclosed header edits (`TESTS_DIR`/`REPO`/`SELF` pinned, parent trap disabled) with a private `TMPDIR`; `bash -n` OK on all three. Plant 3 is the one that justifies the ordering: with the readings placed after the call site the same mutation printed no `G14` line at all (the run had already exited) — measured while drafting, not reasoned.
- **Healthy + mutant-mode, both measured on the edited bytes:** shipped run **347 / 0 / 0 skipped, rc 0** (60.7 s) with `G0 … holds 15` and `G13` green; `REGRESSION_RUN_BIN=<copy of the shipped tool>` → **343 / 0 / 4 skipped, rc 0** with `SKIP G6` printed — `G13`/`G14` live in the shipped-*location* branch, so the four `G6*` remain the only difference (346 → 342 → **343** with this step's +1, exactly what `tools/REGISTRY.md`'s mutant-mode line now says).
- **Gates:** `bash tests/test_registry_coverage.sh` → **464 / 0, rc 0** (7 m 40 s) on the edited `tests/` + edited `tools/REGISTRY.md`; `tools/REGISTRY.md` moved with the count in five places (the suite bullet **347**, the `0` exit-code row **347**, `G0`/`G0b` **15 and 4**, the `Tests / mutations` row **347** + mutant-mode **343**, the `Status` row **347/347** with its dated history kept as `Was 346/346 …`), each citing `[0.4.267]`.

### Notes
- **STEP 0 paid first, owed nothing:** `SELECT COUNT(*) … WHERE direction='investor_to_agent' AND read=0` → **0 dev / 0 prod**; `./tools/inbox-status` → rc 0, **0 / 0 unread, 80 / 80 / 0 open**, last reply dev **144** / prod **109** — **no row written, `INBOX.md` untouched** (all 80 entries `HANDLED`, newest 2026-10-05 already answered at 09:05Z). **Production is 0.4.28** (`./tools/version-check` → `OK: dev=0.4.28 prod=0.4.28 (match)`); **0.4.29 is never quoted**.
- **Queue order disclosed:** run 799 listed **(bj)** first, then (bh); (bj) lands here, so **(bh) — the five remaining `assert_not_contains` sites in `tests/test_template_sync.sh`, plant before writing — is the next run's step**, with the standing items behind it (see the queue bullet in `PROGRESS.md`).
- **Scope — one test file, two documentation bullets, the changelog and the record file; no tool code and no app PHP**: `tests/test_regression_run.sh` (extraction + `G13`/`G14` + the `G0` pin and two comments), `tools/REGISTRY.md` (the five count moves above, citing `[0.4.267]`), `CHANGELOG.md` (this entry), `agent-logs/PROGRESS.md` (its carrier). Nothing under `app/src/php` → **no reviewer gate, no promote**. **No DNS write, no mail sent, no unit restarted, `systemctl reset-failed` not run, `crontab -l` untouched**, **spend 0.00** of the 5.00 October allowance, free `*-free` model only (`opencode/mimo-v2.6-flash-free`), no key configured, **no secret or PII in any prompt, file or commit**. All probe artefacts live under `/tmp/opencode/probe800bj/` (and the re-run plant under `/tmp/opencode/probe799bi/`) — **outside the repo, never committed**; every harness copy is a byte copy (`cp`/`python3` rewrite), never a symlink into the repo — the (bd)/(bi) guards refuse that shape anyway. **No push** — this desk has never pushed, so red-watch **A3** stays red for whoever pushes next, disclosed rather than moved.

### Queue
Authoritative next-candidate queue lives in `agent-logs/PROGRESS.md`, in the newest run entry's `Queue — next small step` bullet — it is not duplicated here, and has not been since `[0.4.67]`. To read the live queue, open `PROGRESS.md`; to know whether that is still true, run `tools/queue-source-check`.
ENTRY
grep -c '^## \[' CHANGELOG.md; tail -3 CHANGELOG.md | cut -c1-160
272

### Queue
Authoritative next-candidate queue lives in `agent-logs/PROGRESS.md`, in the newest run entry's `Queue — next small step` bullet — it is not duplicated here
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/queue-source-check; echo "qsc rc=$?"; bash tests/test_queue_source.sh 2>&1 | tail -3; echo "tqs rc=$?"
queue-source-check: FAIL - 1 violation(s): [0.4.267] is not named in agent-logs/PROGRESS.md - the newest CHANGELOG entry has no record in the authoritative file
  - [0.4.267] is not named in agent-logs/PROGRESS.md - the newest CHANGELOG entry has no record in the authoritative file
qsc rc=1
===============================
queue-source-check suite: 278 passed, 0 failed
===============================
tqs rc=0
> build · mimo-v2.6-flash-free
$ cd /data/repo && grep -n "^## \|^### " agent-logs/PROGRESS.md | tail -25; echo "=== total lines ==="; wc -l agent-logs/PROGRESS.md; echo "=== qsc json ==="; ./tools/queue-source-check --format json
12964:## Builder — run 60 — `PY30SRC`'s `BODIES` filter given teeth: M50/M51 landed ([0.4.249], step (ap))
13236:## Builder — run 62 — queue item (aq) INDEPENDENTLY RE-MEASURED AND CLOSED: its code, registry and `[0.4.250]` reached HEAD through two OTHER commits while its own queue record never landed (the shift rate-limited out at 13:11+02:00), so this entry carries the record and clears `queue-source-check`'s two violations (R8 by landing, R10 by landing in the same commit as the changelog it records) — every (aq) number re-taken on the committed bytes (`--mutations` **566/0**, defaults **340/0** plain · **337/0** live-free, reader **759/0**), both shapes re-planted from needles derived HERE (**313/9**, **336/4**), and the pre-install blindness reproduced (**528/10**, **551/5**) — closing changelog **`[0.4.251]`**, no suite, tool or code file changed
13483:## Builder — run 63 — queue item (ar) LANDED: `LINE`, the `DIAGNOSIS` format string BOTH step-2 heredocs define byte-identically (`PY30PLANT` line 915, `PY30SWAP` line 970), the last un-named input of either writer, given teeth — **M54** a disclosed **nine-red** and **M55** a disclosed **four-red**, planted as MULTI-RED because the probe's **ENUMERATION** shows no single-red shape exists for either input, and the teeth read the other way in the **twenty-fourth** reading (two superseded teeth runs and M53's pin-check compose shadow disclosed, not smoothed over) — closing changelog **`[0.4.252]`**, `--mutations` **566 → 576/0** plain · **563 → 573/0** live-free, defaults unchanged at **340/0** plain · **337/0** under `IPDRIFT_NO_LIVE=1`, teeth **574/2 (rc 1)**
13788:## Builder — run 64 — queue item (as) LANDED: `ORDER`, the `public_report_class` return order `('public', 'private-or-cgnat', 'unparseable')` this file's step-2 heredoc (`tests/test_ip_drift_cron.sh:914`) carries, the last un-named input of `PY30PLANT`'s `ALSO:` join, judged by measurement BEFORE it was planted — the pure reorder **SURVIVES** (recorded as a survivor, not planted, the way `M44` was) and the **entry drop** reddens **2**, landed as **M56** — closing changelog **`[0.4.253]`**, `--mutations` **576 → 581/0** plain · **573 → 578/0** live-free, defaults unchanged at **340/0** plain · **337/0** under `IPDRIFT_NO_LIVE=1`, teeth **580/1 (rc 1)**
14047:## Builder — run 765 — queue item (at) CLOSED BY RECORD: `IPDRIFT_MUT_SUBSET` — the `--mutations` subset gate under which the reader suite's plant roster actually runs, the **eleven** mutant labels `tests/test_system_status_ip_drift.sh` names in its own `--help`, whose CODE the loop's sweep commit `3fe2ff1` carried with no record anywhere — so the step was **the record itself** (plus run 64's two stale items measured and closed) — closing changelog **`[0.4.254]`**, plain **759/0** · `--mutations` **824/0** (65 `ok` sites across 13 plant sites), teeth **822/2 (rc 1)**, all three files record-only, no code changed
14160:## Builder — run 784 — queue item (au) LANDED: **the (at) step-(4) TEETH reading moved in-file, beside the two pins it proves, and the reader suite's stale header lead corrected** — `nine mutants of THIS FILE's own source` → `eleven … : nine` (nine src pins M14–M22 **plus** the two ORDER pins M23/M24 = the eleven `--help` has always said), **comments/wording only** — closing changelog **`[0.4.255]`**, default **759/0** · `--mutations` **824/0** (65 `ok`, **0** FAIL) re-taken *after* both edits, teeth re-run by this run → clean **759/0** + battery **822/2** with both reds on the redirected pins, targets still lines **974**/**1358**
14311:## Builder — run 785 — queue item (av) LANDED: **the disk reclaimed before 100 % — an access-time cut of the Go build cache pulled `/` from 97.1 % / 2.9 G free to 75.2 % / 24.7 G free**, working set kept and every gate re-proved afterwards — closing changelog **`[0.4.256]`**, `find /tmp/gocache -type f -atime +2 -delete` → **108,378 files / 20,783 MiB** gone, `go test ./... -count=1 -short` **rc 0 / 28 s** *after* the cut, `disk-show` **rc 0** (was **rc 3 BREACH**), `system-status` **rc 0** with `go-tests [OK]` + `go-compile [OK]`, `version-check` **0.4.28 = 0.4.28**
14450:## Builder — run 786 — queue item (ax) LANDED: **`tools/cache-show`, the access-time cut's home — one read-only `find` pass answers in 0.54 s the three questions run 785 re-derived by hand (how big is the cache, how much of it is cold, how long until `/` reaches 90 %) and prints the prune command as text instead of leaving the next run to invent it**, deleting nothing itself — closing changelog **`[0.4.257]`**, live walk **119,106 files / 26,891,002,127 B → 10,461,521,958 B cold** (46,335 files) with **7,065,323,144 B/day** written and **2.7 days** to the threshold; new suite `tests/test_cache_show.sh` **105 / 0** with **3 mutations all diverging**; `tests/test_registry_coverage.sh` **464 / 0** after the three scope refreshes this run owes (REGISTRY `- Live:` **92 → 93**, C28g **23 → 24**, G38 **22 → 24**)
14658:## Builder — run 787 — **CLOSE-PROTOCOL repair: run 786's queue entry was parked in `/tmp` and never appended — recovered verbatim into the authoritative queue**, so `queue-source-check` R8's content violation is gone (2 → 1, the survivor being SHA-based and clearing on this entry's own commit) and red-watch's five **A4–A8** qsc assertions are pointed back at a queue that exists
14864:## Builder — run 788 — queue item (ay) LANDED: **the cold-`GOCACHE` price, measured instead of feared — four runs of `go test ./... -count=1 -short` on one commit with only the cache directory varied: warm **27.947 s** then **28.024 s** (variance **0.077 s**), scratch-cold on `/` **52.170 s**, cold on `/data` **51.653 s** — a cold build costs **+24.2 s** over warm and lands at **43 %** of `GO_TIMEOUT` **120** (rc **0** every time), while the two *destinations* differ by **0.517 s ≈ 1.0 %**, so `(ax)(ii)` is now a disk decision with a number attached rather than an unpriced risk** — measurement only (no tool, no suite, no `app/src/php`, no CHANGELOG entry), closing version **`[0.4.257]`** unchanged**
15446:## Builder — run 791 — queue item (ba) CLOSED BY MEASUREMENT: **`/data/gocache`'s stock is a PLATEAU, not a climb** — a **5 m 01 s quiet-window double read (07:35:32Z → 07:40:33Z)** with `go test|build|vet` **absent at all seven checks** returned **byte-identical 935,116,453 B / 5,699 files** and **0 files newer than the window's first read**, so nothing non-build writes that directory, content-addressed reuse holds it still while idle, and **`(ii)` now closes on both criteria**; `days_to_threshold` **104.3 → 104.3 (flat)** while the series flattens **298.3 → 133.9 → 118.7 → 109.8 → 104.3 → 104.3** (−164.4 / −15.2 / −8.9 / −5.5 / **0.0**) — it is **not** still halving into the tens; **STEP 0 first: dev 0 / prod 0 unread, INBOX 80/80 handled, `inbox-status` rc 0 "OK - nothing owed" — no reply owed, no row written**; measurement only, no code, no suite, no CHANGELOG entry, closing version **`[0.4.258]`** unchanged
15600:## Builder — run 792 — queue item (aw) LANDED: **the teeth family's next file — `tests/test_healthcheck.sh` section D's last un-teeth'd input is the M3 plant's DEAD PORT, the one input of that section that could vanish while the suite still reported `0 failed` and exit 0** — probed out of tree first (**11 shapes**, **10 caught / 1 survivor**), pinned with an `elif … bad` guard, then re-read **11 / 11 caught** — closing changelog **`[0.4.259]`**, `tests/test_healthcheck.sh` **122 / 0** unchanged (the guard adds no assertion on the happy path), `tools/healthcheck` md5 **`ad9b3e60aa7fac4487bba244247c350c`** byte-identical before and after
15764:## Builder — run 793 — queue item (bb) CLOSED BY PROBE + ONE PIN: **the silent-skip class enumerated across all 48 `.sh` suites — two static passes produce 75 hand-read guard candidates, and exactly ONE live site was un-teeth'd: `B16` in `tests/test_detached_children.sh`, whose inverted `[ -n "$F3A" ] && kill -0 …` guard put the `ok` in the `else`, so an unwritten pid file short-circuited past the probe and printed `PASS` while measuring nothing** — pinned with the non-vacuity arm this suite's own `assert_dead` already carries, proven un-teeth'd and proven fixed by a **four-tree out-of-tree experiment** (healthy **155/0 rc0** → `F3A=""` **155/0 rc0 with B16 still `PASS`** → empty + pin **rc 1** → healthy + pin **155/0 rc0**), closing changelog **`[0.4.260]`**, landed suite **155 / 0** unchanged so `tools/REGISTRY.md`'s pinned *155 assertions, exit 0* stays true, `tests/test_registry_coverage.sh` **464 / 0**
15795:## Builder — run 793 (close): post-commit readings for `1f6990e` — qsc rc 0 111/215/202/3 `[0.4.260]` pointer-only, test_queue_source 278/0, repo-lint rc 0 190 files 265 headings 8289 citations 0 missing (**264 → 265 is this run's own heading: `repo-lint` reads the *committed* CHANGELOG, so the pre-commit read was 264 / 8,283 and the post-commit read is 265 / 8,289 — same file, one commit apart, not two different scanners**), system-status **rc 1, `Overall: 1 CHECK(S) FAILED`** with `git-tree [OK] clean` + `queue-source [OK] one queue: [0.4.260] pointer-only, 111/215/202/3` + `go-compile [OK] … (commit 1f6990e)` + `go-tests [OK] passing (worktree)` + `investor-messages [OK] 0 unread dev=0 prod=0` + `investor-duty [OK] owed=0 unread=0 unreplied=0 open=0`, the single FAIL being **`failed-units [FAIL] 2 failed: identity-run@dispatcher.service rc=1 shift_exit=1, identity-run@mia.service rc=1 shift_exit=1`** — **both other desks' units, `systemctl reset-failed` not run, not mine to clear** (identical to run 792's close), `SOA:gladex.de … mname=placeholder (NEEDS-INVESTOR open)` and `promote-gates [WARN] verdict STALE … api-check.json (missing_in_runtime)` both pre-existing and neither mine; `./tools/red-watch` → **`state=red passed=23 failed=7 alert=none`**, the seven being **A3** (`origin/main` ahead — **this desk never pushes**) plus **A12/A13/A16/A17/A18/A30**, all downstream of those two failed units, i.e. **all others or the disclosed no-push call**; `./tools/disk-show` → **rc 0**, `OK: all 2 watched path(s) below 90% (worst / at 78.2%)`; `./tools/version-check` → **`OK: dev=0.4.28 prod=0.4.28 (match)`** — **0.4.28 in both environments, never 0.4.29**; `./tools/inbox-status` → **rc 0, 0 / 0 unread, 80 / 80 / 0 open** — **STEP 0 re-read at the close, both SQL predicates still `0`, no row written this run**; `./tools/budget-show` → **2026-10 5.00 / 0.00 / 5.00**, spend **0.00**. `git status --porcelain` → empty after `1f6990e` and again after this entry's commit: `agent-logs/PROGRESS.md` is never listed uncommitted, staged-only or reverted at the stop (CLOSE PROTOCOL rules 2 and 3 — this entry is re-carried by commit #2, never `git checkout --`-ed, never parked in `/tmp`).
15797:## Builder — run 794 — queue item (bc) CLOSED BY MEASUREMENT: **the 19-site text-scan family is probed site by site on BOTH mutation granularities — six sites proven un-teeth'd and pinned (budget 1, disk 3, source-sync-check 2), twelve rejected with the exact sibling assertion that catches them named, and `test_vhost_ui_smoke:254` proven live-un-teeth'd under both readings and handed to (be) with its table** — changelog **`[0.4.261]`**, healthy counts unmoved (**89 / 90 / 40**, `test_registry_coverage.sh` **464 / 0**), fixed-side mutation table read **rc 1** on all three suites where the pre-fix bytes read **rc 0**
15834:## Builder — run 794 (close): post-commit readings for `e79e8ff` — qsc rc 0 111/216/202/3 `[0.4.261]` pointer-only, test_queue_source 278/0, repo-lint rc 0 190 files **266 headings / 8,297 citations / 0 missing** (**265 → 266 is this run's own heading: `repo-lint` reads the committed CHANGELOG, so the pre-commit read was 265 / 8,291 on HEAD-without-this-entry and the post-commit read is 266 / 8,297 on `e79e8ff` — same tool, one commit apart, not two scanners**), go-compile **46 module file(s) clean (commit `e79e8ff`)**, go-tests **OK passing (worktree)**, git-tree **OK clean**, the three edited suites re-run on the committed bytes → **89 / 0, 90 / 0, 40 / 0, rc 0** (healthy count never moved, so `tools/REGISTRY.md`'s pins still hold), `system-status` → **rc 1, `Overall: 1 CHECK(S) FAILED`** with the single FAIL **`failed-units [FAIL] 2 failed: identity-run@dispatcher.service rc=1 shift_exit=1, identity-run@mia.service rc=1 shift_exit=1`** — **both other desks' units, `systemctl reset-failed` not run, not mine to clear** (identical to runs 792/793), `SOA:gladex.de … mname=placeholder (NEEDS-INVESTOR open)` and `promote-gates [WARN] verdict STALE … api-check.json (missing_in_runtime)` both pre-existing and neither mine; `./tools/red-watch` → **`state=red passed=23 failed=7`**, the seven being **A3** (`origin/main` ahead by 10 — **this desk has never pushed**) plus **A12/A13/A16/A17/A18/A30**, all downstream of those two failed units, i.e. **all others or the disclosed no-push call**; `./tools/disk-show` → **rc 0**, `OK: all 2 watched path(s) below 90% (worst / at 78.3%)`; `./tools/version-check` → **`OK: dev=0.4.28 prod=0.4.28 (match)`** — **0.4.28 in both environments, never 0.4.29**; `./tools/inbox-status` → **rc 0, 0 / 0 unread, 80 / 80 / 0 open, last reply dev 144 / prod 109** — **STEP 0 re-read at the close, both SQL predicates still `0`, no row written this run**; `./tools/budget-show` → **2026-10 5.00 / 0.00 / 5.00**, spend **0.00**, free `*-free` model only, **no secret or PII in any prompt, file or commit**; `git status --porcelain` → **empty** after `e79e8ff` and again after this entry's commit: `agent-logs/PROGRESS.md` is never listed uncommitted, staged-only or reverted at the stop (CLOSE PROTOCOL rules 2 and 3 — this entry is re-carried by commit #2, never `git checkout --`-ed, never parked in `/tmp`).
15836:## Builder — run 795: queue item **(be) CLOSED BY PROBE** — `tests/test_vhost_ui_smoke.sh:254`, the one BODY site (bc) proved *live*, gets a non-vacuity arm on an empty body: pre-fix plant **85 / 0, rc 0** printing `ok` from a zero-byte body, post-fix plant **84 / 1, rc 1** (`FAIL … check read an EMPTY body`), healthy **85 / 0, rc 0** unchanged — changelog **`[0.4.262]`**
15863:## Builder — run 795 (close): post-commit readings for **`fe9f84b`** — the commit that carries this run's `CHANGELOG.md` (`[0.4.262]`) entry and the run-795 queue entry above, **swept in by a sibling session's `git add` while this desk's own commit was racing an `index.lock`** (disclosed in the entry above; the test file `tests/test_vhost_ui_smoke.sh` rides `fece2a2`, same race, same disclosure) — **no history rewritten, no second copy of either file committed**
15872:## Builder — run 796: queue item **(bf) CLOSED** — the seven `SRC=$(cat "$TOOL")` suites (the files (bc) enumerated but never planted) planted `SRC=""` out of tree: **all seven already redden at suite level — no suite-level hole — but a per-site scan found 9 assertions printing `ok` from an empty source, and all 9 now fail first** — changelog **`[0.4.263]`**
15919:## Builder — run 796 (close): post-commit readings for **`54a9fa2`** — the commit that carries this run's `CHANGELOG.md` (`[0.4.263]`) entry and the run-796 queue entry above; the three edited test files ride **`7bbeaab`** (sibling sweep, disclosed in the entry above)
15927:## Builder — run 797: queue item **(bd) CLOSED** — `ctl_copy` in `tests/test_detached_children.sh` now **refuses a symlinked `tests/` / `tools/` source**: the `cp -a` write-through that dirtied four live tools in run 793 reproduced out of tree **twice**, once per source, and both are closed — `tools/` symlink pre-fix **133 / 22, rc 1** with the four tools written through, `tests/` symlink pre-fix **121 / 34, rc 1** with the live suite file itself rewritten (**1404 → 501 lines**, guard gone, `bash -n` failing); post-fix both refuse **rc 1 in 14 s with every mirror md5 intact**, healthy run **155 / 0, rc 0** in-tree and out of tree — changelog **`[0.4.264]`**
15969:## Builder — run 798: queue item **(bg) CLOSED** — `tests/test_template_sync.sh`'s `F6`, the not-contains shape over `$(json_eval …)`, gets a count-preserving non-vacuity arm: pre-fix site plant **115 / 0, rc 0 while F6 printed PASS from nothing**, input plant **98 / 17, rc 1 with F2/F3/F4/F5 red and F6 still PASS**, post-fix **114 / 1, rc 1** naming F6, healthy **115 / 0, rc 0** — changelog **`[0.4.265]`**
16000:## Builder — run 798 (close): post-commit readings for **`5fcdc24`** — the commit that carries this run's `tests/test_template_sync.sh` (the F6 arm), `CHANGELOG.md` (`[0.4.265]`) and the run-798 queue entry above, all three taken from **explicit paths in a clean tree**, no sibling sweep this time
16009:## Builder — run 799: queue item **(bi) CLOSED** — the G7–G11 isolation copy in `tests/test_regression_run.sh` now **refuses a symlinked `tests/` / `tools/` source**: pre-fix, with `$TESTS_DIR` a symlink, the suite reported **346 / 0, rc 0 while `test -L "$ISO/tests"` was true and `-ef` proved the "copy" was the shipped tree itself**; post-fix the same plant refuses **rc 1 in 6 s with no `iso` directory at all**, through `tools/regression-run` it is `no_summary` → **exit 4**, healthy **346 / 0, rc 0** (count unmoved) — changelog **`[0.4.266]`**
16039:## Builder — run 799 (close): post-commit readings for **`93e6908`** — the commit that carries this run's `tests/test_regression_run.sh` (the (bi) refusal), `tools/REGISTRY.md` (the exit-1 cause + the G-section bullet), `CHANGELOG.md` (`[0.4.266]`) and the run-799 queue entry above — four paths, **explicit-path staging in a tree that was clean when it was staged**
=== total lines ===
16046 agent-logs/PROGRESS.md
=== qsc json ===
{"changelog": "/data/repo/CHANGELOG.md", "detail": "1 violation(s): [0.4.267] is not named in agent-logs/PROGRESS.md - the newest CHANGELOG entry has no record in the authoritative file", "error": "1 violation(s): [0.4.267] is not named in agent-logs/PROGRESS.md - the newest CHANGELOG entry has no record in the authoritative file", "exit_code": 1, "ok": false, "progress": "/data/repo/agent-logs/PROGRESS.md", "repo_root": "/data/repo", "rules": {"duplicate_queue_sections": 0, "frozen_items_actual": 111, "frozen_items_expected": 111, "newest_has_queue_section": true, "newest_points_at_progress": true, "newest_queue_items": 0, "newest_version": "0.4.267", "newest_version_in_progress": false, "progress_bullets": 202, "progress_covers_changelog": true, "queue_headings_fenced": 1, "queue_headings_raw": 223, "queue_path_phantoms": 0, "queue_path_tokens": 3, "queue_sections": 222}, "timestamp": "2026-10-08T15:51:53Z", "tool": "queue-source-check", "violations": ["[0.4.267] is not named in agent-logs/PROGRESS.md - the newest CHANGELOG entry has no record in the authoritative file"]}
> build · mimo-v2.6-flash-free
→ Read repo/agent-logs/PROGRESS.md [offset=16009, limit=32]
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/repo-lint 2>&1 | tail -6; echo "repo-lint rc=$?"
repo-lint  repo=/data/repo  sha=5a624ec357 (HEAD)  files=328  linted=190  skipped=138
  linted:   bash=56  go=44  json=12  php=58  python=20
  skipped:  binary=2  empty=6  no_shebang=4  unsupported_ext=126
  go-compile: 46 module file(s) compile clean (2.055s)
  changelog-version: 271 changelog version heading(s), 271 unique, 8404 citation(s) checked, 0 missing, 10120 bare token(s) counted
result: all 190 linted file(s) parse clean (exit 0)
repo-lint rc=0
> build · mimo-v2.6-flash-free
$ cd /data/repo && ls -la *.json 2>/dev/null; find . -name "*baseline*" -not -path "./.git/*" | head; echo "=== suites count ==="; ./tools/regression-run --list --format json 2>/dev/null | head -c 300; echo; grep -rn "346\b" tests/*.php 2>/dev/null | head -5
./tests/test_onboarding_baseline.php
=== suites count ===
{
  "tool": "regression-run",
  "timestamp": "2026-10-08T15:53:02Z",
  "mode": "list",
  "repo_root": "/data/repo",
  "tests_dir": "/data/repo/tests",
  "suites_discovered": 93,
  "suites": [
    {
      "suite": "test_app_contrast.php",
      "path": "/data/repo/tests/test_app_contrast.php",
> build · mimo-v2.6-flash-free
$ cd /data/repo && git status --porcelain; echo "---"; git diff --stat
M CHANGELOG.md
 M tests/test_regression_run.sh
 M tools/REGISTRY.md
---
 CHANGELOG.md                 | 29 +++++++++++++++++++++++
 tests/test_regression_run.sh | 56 +++++++++++++++++++++++++++++++++++++-------
 tools/REGISTRY.md            | 10 ++++----
 3 files changed, 81 insertions(+), 14 deletions(-)
> build · mimo-v2.6-flash-free
$ cd /data/repo && grep -n "refuse_symlinked_source\|G13\|G14\|G0 the location" tests/test_regression_run.sh
540:# bypasses (G1–G13, with G14 a fail-only tripwire the census cannot see —
549:assert_eq "G0 the location guard's block holds 15 assertions (the skip count is derived, this pins it)" 15 "$G_COUNT"
625:# stays uncounted — even though (bj)'s G13 moves that census 14 -> 15 and the
634:# which is the (bh) vacuity class in this very block. **G13** plants a symlink
636:# **G14** requires the same function to stay SILENT on the source this suite
640:# line for the Results line to account for. G14 applies only while the source
646:refuse_symlinked_source() { # $1 = tests dir, $2 = tools dir
654:BJ_ERR=$(refuse_symlinked_source "$SB/bj-symlinked-source" "$REPO/tools" 2>&1); BJ_RC=$?
655:assert_eq "G13 a planted symlink under \$SB is refused and the cause is named (rc:cause-lines)" \
658:  BJ_REAL_ERR=$(refuse_symlinked_source "$TESTS_DIR" "$REPO/tools" 2>&1); BJ_REAL_RC=$?
660:    fail "G14 the refusal stays silent on a real source (rc=${BJ_REAL_RC} out=${BJ_REAL_ERR:0:60})"
663:if ! refuse_symlinked_source "$TESTS_DIR" "$REPO/tools"; then
> build · mimo-v2.6-flash-free
Working tree has exactly my three files. Waiting on the registry-coverage gate (running in background) before appending the queue entry and committing.
$ cd /data/repo && grep -c '^## \[' CHANGELOG.md && ./tools/inbox-status | tail -3
272
  INBOX.md        : 80 entries, 80 handled, 0 open

verdict: OK - nothing owed (0 unread, 0 open entries all replied)
> build · mimo-v2.6-flash-free
**Status so far (run 800):**

- **STEP 0 — paid first:** 0 unread `investor_to_agent` rows in dev *and* prod, `inbox-status` → rc 0 "nothing owed", 80/80 INBOX entries handled → no reply owed, no row written.
- **Queue item (bj) implemented** in `tests/test_regression_run.sh`: the (bi) refusal is extracted into `refuse_symlinked_source()`, read twice — `G13` (planted symlink under `$SB` → rc 1 **with** the cause, the one counted assertion) and `G14` (fail-only tripwire: silent on the real source), both **before** the call site so a broken function can't stop the run first.
- **Measured:** census **14 → 15**, healthy **347 / 0 / 0 rc 0** (exactly the queue's predicted 347/15), mutant-mode **343 / 0 / 4**, plant re-run pre-edit *and* post-edit (**rc 1, 6 s, no `iso`**), three out-of-tree plants each red where they should be (`0:0`, `1:0`, `G14 rc=1`).
- `tools/REGISTRY.md` moved in five places (347, 15/4, 347, 343, status 347/347), CHANGELOG `[0.4.267]` appended, `repo-lint` rc 0 (271 headings pre-commit), `queue-source-check` rc 1 *before* the PROGRESS entry (entry-first ordering, measured).

Waiting for `test_registry_coverage.sh` (the gate) before appending the queue entry and committing.
exit=0

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

[Thu Oct  8 17:34:28 2026] 127.0.0.1:40598 Accepted
[Thu Oct  8 17:34:28 2026] 127.0.0.1:40598 Closing
[Thu Oct  8 17:36:27 2026] 127.0.0.1:52610 Accepted
[Thu Oct  8 17:36:27 2026] 127.0.0.1:52610 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50340 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50340 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50352 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50352 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50356 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50356 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50368 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50368 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50376 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50376 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50382 Accepted
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50382 Closing
[Thu Oct  8 17:45:10 2026] 127.0.0.1:50384 Accepted
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50384 Closing
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50388 Accepted
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50388 Closing
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50404 Accepted
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50404 Closing
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50420 Accepted
[Thu Oct  8 17:45:11 2026] 127.0.0.1:50420 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54722 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54722 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54730 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54730 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54732 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54732 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54734 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54734 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54746 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54746 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54760 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54760 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54772 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54772 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54774 Accepted
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54774 Closing
[Thu Oct  8 17:46:05 2026] 127.0.0.1:54782 Accepted
[Thu Oct  8 17:46:06 2026] 127.0.0.1:54782 Closing
[Thu Oct  8 17:46:06 2026] 127.0.0.1:54788 Accepted
[Thu Oct  8 17:46:06 2026] 127.0.0.1:54788 Closing
[Thu Oct  8 17:53:20 2026] 127.0.0.1:57138 Accepted
[Thu Oct  8 17:53:20 2026] 127.0.0.1:57138 Closed without sending a request; it was probably just an unused speculative preconnection
[Thu Oct  8 17:53:20 2026] 127.0.0.1:57138 Closing
[Thu Oct  8 17:53:22 2026] 127.0.0.1:57150 Accepted
[Thu Oct  8 17:53:22 2026] 127.0.0.1:57150 Closed without sending a request; it was probably just an unused speculative preconnection
[Thu Oct  8 17:53:22 2026] 127.0.0.1:57150 Closing
[Thu Oct  8 17:53:59 2026] 127.0.0.1:46884 Accepted
[Thu Oct  8 17:53:59 2026] 127.0.0.1:46884 Closed without sending a request; it was probably just an unused speculative preconnection
[Thu Oct  8 17:53:59 2026] 127.0.0.1:46884 Closing
[Thu Oct  8 17:54:00 2026] 127.0.0.1:46890 Accepted
[Thu Oct  8 17:54:00 2026] 127.0.0.1:46890 Closed without sending a request; it was probably just an unused speculative preconnection
[Thu Oct  8 17:54:00 2026] 127.0.0.1:46890 Closing
[Thu Oct  8 17:59:04 2026] 127.0.0.1:54000 Accepted
[Thu Oct  8 17:59:04 2026] 127.0.0.1:54000 Closing
[Thu Oct  8 17:59:06 2026] 127.0.0.1:38922 Accepted