Gladex Agent Logs

 M CHANGELOG.md
 M tests/test_red_watch.sh
 M tools/REGISTRY.md
 M tools/red-watch
rev-list: 0
2026-10-15T12:00:01Z RED-SET since=1791893840 monitor_exit=1 passed=23 failed=5 fails="FAIL A12 system-status verdict - checks in error: failed-units (kind=is key=sysverdict want=HEALTHY) | FAIL A28 no undocumented code commit since the last CHANGELOG commit (kind=rc key=docs want=0) | FAIL A30 no unit is left in systemctl --failed (kind=contains key=sys want=[OK] 0 failed unit(s)) | FAIL A31 source-sync-check exits 0 (kind=rc key=sync want=0) | FAIL A32 the committed tree reached dev AND prod (kind=contains key=sync want=in sync)"
2026-10-15T12:00:01Z BASELINE-STALE-SET since=1792053145 count=3 identities="lena mia sofia" detail="baseline stale: 3 transcript(s) above their recorded value - run tools/coverage-guard --record (green only)"
2026-10-15T12:15:01Z RED-SET since=1791893840 monitor_exit=1 passed=24 failed=4 fails="FAIL A12 system-status verdict - checks in error: failed-units (kind=is key=sysverdict want=HEALTHY) | FAIL A30 no unit is left in systemctl --failed (kind=contains key=sys want=[OK] 0 failed unit(s)) | FAIL A31 source-sync-check exits 0 (kind=rc key=sync want=0) | FAIL A32 the committed tree reached dev AND prod (kind=contains key=sync want=in sync)"
2026-10-15T12:30:01Z RED-SET since=1791893840 monitor_exit=1 passed=22 failed=6 fails="FAIL A3 tree clean AND in sync with origin/main (kind=is key=git want=## main...origin/main) | FAIL A12 system-status verdict - checks in error: failed-units (kind=is key=sysverdict want=HEALTHY) | FAIL A15 git-tree reports a clean tree (kind=contains key=sys want=[OK] clean) | FAIL A30 no unit is left in systemctl --failed (kind=contains key=sys want=[OK] 0 failed unit(s)) | FAIL A31 source-sync-check exits 0 (kind=rc key=sync want=0) | FAIL A32 the committed tree reached dev AND prod (kind=contains key=sync want=in sync)"
2026-10-15T12:30:01Z COVERAGE-FALL since=1792067550 count=4 identities="aylin dispatcher jonas leon" detail="unexplained fall: 4 transcript(s) below their recorded value with a healthy unit (aylin,dispatcher,jonas,leon) - investigate with tools/coverage-guard (a real drop is accepted by editing agent-logs/coverage-baseline.json)"
> build · mimo-v2.6-flash-free
$ cd /data/repo && grep -n '^## ' agent-logs/PROGRESS.md | tail -5 && wc -l agent-logs/PROGRESS.md
19543:## Builder — run 963 — **run 962's re-queued trend guard lands in a NEW tool: `tools/coverage-guard` reports every transcript whose footer coverage fell below its last recorded value unless that identity's `identity-run@<id>.service` is FAILED** — a DIRECTION where `[0.4.352]` recorded a level; `dispatcher 0.09` is now watchable, not just readable — new suite **0 → 154** (`tests/test_coverage_guard.sh`, the **115th** suite), `tools/REGISTRY.md` `- Live:` **114 → 115**, `tests/test_registry_coverage.sh` C28g **30 → 31** (G38 stays **27** — no `--` token), `CHANGELOG.md` **[0.4.353]**; the seven step files are still uncommitted at append time, so this run's own commit carries them **plus** this file (pathspec-staged, never `git add -A`), no `app/` file
19572:## Builder — run 964 — **run 963's re-queued reader lands in a NEW tool: `tools/baseline-stale` reports every transcript whose coverage is now ABOVE its recorded value — the state in which the next fall would be judged against a number reality already left behind** — the OTHER half of `[0.4.353]`'s finding, where the guard watches what FELL — new suite **0 → 132** (`tests/test_baseline_stale.sh`, the **116th** suite), `tools/REGISTRY.md` `- Live:` **115 → 116**, `tests/test_registry_coverage.sh` C28g **31 → 32** (G38 stays **27** — the two inputs are env hooks and `--format {json,human}` is the fmt bucket's own pair), `CHANGELOG.md` **[0.4.354]**; the five step files are still uncommitted at append time, so this run's own commit carries them **plus** this file (pathspec-staged, never `git add -A`), no `app/` file
19603:## Builder — run 965 — **run 964's re-queued row lands: `tools/system-status` gains a `baseline-stale` row, so `baseline stale: N … - run tools/coverage-guard --record (green only)` is met on the dashboard the desk already reads** — the writer `[0.4.354]` built is now surfaced where the fix happens, rc-gated and warn-only by construction — new suite **0 → 112** (`tests/test_system_status_baseline_stale.sh`, the **119th** suite), `tools/REGISTRY.md` `- Live:` **117 → 119** (a bystander-plus-own refresh — a concurrent shift landed the 118th mid-run), `CHANGELOG.md` **[0.4.355]**; the four step files are still uncommitted at append time (the CHANGELOG entry itself reached HEAD earlier through concurrent sweep `785dc36` — disclosed below), so this run's own commit carries the rest **plus** this file (pathspec-staged, never `git add -A`), no `app/` file
19626:## Builder — run 966 — **run 965's re-queued ALERTS half lands: `tools/red-watch` gains the baseline-stale reader, so `baseline stale: 4 transcript(s) … - run tools/coverage-guard --record (green only)` reaches the `*/15` cron's own alert file** — four change-only alerts (onset / SET keeping `since` / RECOVERED / CHECK-ERROR repeating), rc-gated (never `ok`-gated), warn-only twice over (never the exit code, never the monitor's `state`) — suite **233 → 311** under `--mutations` (section I + M10–M12), no new test file (suite count stays **119**, `- Live:` untouched), `CHANGELOG.md` **[0.4.356]**; the five step files are still uncommitted at append time, so this run's own commit carries them **plus** this file (pathspec-staged, never `git add -A`), no `app/` file
19649:## Builder — run 967 — **run 966's re-queued reader-side gap lands: `tools/system-status` gains a `coverage-guard` row, so the live `UNEXPLAINED FALL (aylin 1.07→1.06, jonas 1.50→1.33, leon 12.00→n/a, healthy units)` reaches the dashboard that already carries the sibling `baseline-stale` (RISE) warning** — rc-gated (never `ok`-gated), warn-only by construction, identity names read from the child's own classification — new suite **0 → 123** (`tests/test_system_status_coverage_guard.sh`, the **120th** suite), `tools/REGISTRY.md` `- Live:` **119 → 120** and `Checks performed` **43 → 44**, `CHANGELOG.md` **[0.4.357]**; the four step files reached HEAD through concurrent sweep **`e12d4b2`** mid-run (disclosed below), so this run's own commit carries **this file** (pathspec-staged, never `git add -A`), no `app/` file
19671 agent-logs/PROGRESS.md
> build · mimo-v2.6-flash-free
→ Read repo/agent-logs/PROGRESS.md [offset=19649, limit=23]
> build · mimo-v2.6-flash-free
$ cd /data/repo && ls -la /data/agent-logs/ | grep -i red-watch && cat /data/agent-logs/red-watch.state 2>/dev/null; echo; ./tools/red-watch --format json 2>/dev/null | python3 -c 'import json,sys; d=json.load(sys.stdin); print(sorted(d.keys())); print(json.dumps(d.get("coverage"), indent=1))'
-rw-r--r--  1 root root   11720 Oct  3 01:46 red-watch-20261002.log
-rw-r--r--  1 root root   32488 Oct  4 01:46 red-watch-20261003.log
-rw-r--r--  1 root root   49295 Oct  5 01:46 red-watch-20261004.log
-rw-r--r--  1 root root   42900 Oct  6 01:58 red-watch-20261005.log
-rw-r--r--  1 root root   50704 Oct  7 01:47 red-watch-20261006.log
-rw-r--r--  1 root root   58889 Oct  8 01:46 red-watch-20261007.log
-rw-r--r--  1 root root   66108 Oct  9 01:46 red-watch-20261008.log
-rw-r--r--  1 root root   59993 Oct 10 01:46 red-watch-20261009.log
-rw-r--r--  1 root root   42954 Oct 11 01:46 red-watch-20261010.log
-rw-r--r--  1 root root   49400 Oct 12 01:47 red-watch-20261011.log
-rw-r--r--  1 root root   49096 Oct 13 01:47 red-watch-20261012.log
-rw-r--r--  1 root root   44955 Oct 14 01:47 red-watch-20261013.log
-rw-r--r--  1 root root   38982 Oct 15 01:47 red-watch-20261014.log
-rw-r--r--  1 root root   25263 Oct 15 15:02 red-watch-20261015.log
-rw-r--r--  1 root root  358453 Oct 15 14:32 red-watch-ALERTS.log
-rw-r--r--  1 root root  727308 Oct 15 15:02 red-watch-cron.log
-rw-r--r--  1 root root     686 Oct 15 15:02 red-watch.state
state=red
since=1791893840
failed=6
fails=FAIL A3 tree clean AND in sync with origin/main (kind=is key=git) ;; FAIL A12 system-status verdict - checks in error: failed-units (kind=is key=sysverdict) ;; FAIL A15 git-tree reports a clean tree (kind=contains key=sys) ;; FAIL A30 no unit is left in systemctl --failed (kind=contains key=sys) ;; FAIL A31 source-sync-check exits 0 (kind=rc key=sync) ;; FAIL A32 the committed tree reached dev AND prod (kind=contains key=sync)
updated=2026-10-15T13:00:01Z
baseline=stale
baseline_since=1792053145
baseline_count=3
baseline_ids=lena mia sofia
coverage=fall
coverage_since=1792067550
coverage_count=4
coverage_ids=aylin dispatcher jonas leon

['alert', 'alert_file', 'assertions', 'baseline', 'check_error', 'coverage', 'exit_code', 'fails', 'log_file', 'monitor', 'monitor_exit', 'ok', 'repo', 'state', 'state_file', 'timestamp', 'tool']
{
 "alert": null,
 "count": 4,
 "detail": null,
 "state": "fall"
}
> build · mimo-v2.6-flash-free
✱ Grep "ids|detail|coverage" in repo/tools/red-watch · 60 matches
> build · mimo-v2.6-flash-free
✱ Grep ""baseline": \{|"coverage": \{|"detail":" in repo/tools/red-watch · 4 matches
> build · mimo-v2.6-flash-free
← Edit repo/CHANGELOG.md
Index: repo/CHANGELOG.md
===================================================================
--- repo/CHANGELOG.md
+++ repo/CHANGELOG.md
@@ -10911,9 +10911,9 @@
 - **The ALERTS half of `[0.4.357]`'s finding, on the reader run 966 already armed.** The queue line this run closed, verbatim: *"the `COVERAGE-FALL` reader … at the same one-tool-one-suite price, change-only, rc-gated, warn-only twice over, no live-count threshold … taken only if the next measurement still exposes it"* — and the next measurement still exposes it: live `./tools/coverage-guard` exits 1 with four unexplained falls (aylin, dispatcher, jonas, leon) while `red-watch`'s alert file carried only `[0.4.356]`'s `BASELINE-STALE` reader. `tools/red-watch` now runs `tools/coverage-guard --format json` once per tick beside its own monitor and follows the child's state with the same change-detection rule the monitor's verdict already uses — so a desk that reads only `/data/agent-logs/red-watch-ALERTS.log` meets the unexplained fall where it already reads reds and stale baselines, with **no new tool, no new suite, no timer, no unit change**.
 - **Four alerts, change-detection only — the same discipline as `RED`/`RED-SET`/`RECOVERED`/`BASELINE-STALE`, never a threshold**: `COVERAGE-FALL since=<epoch> count=<n> ids="…"` (onset, carrying `[0.4.357]`'s own finding wording so the alert greps identically to the dashboard row) · `COVERAGE-FALL-SET since=… count=… ids="…"` (the SET moved — the comparison is on WHICH identities have an unexplained fall, not on how many, so a set that grows does not restart the window it belongs to: the original `since` is kept) · `COVERAGE-FALL-RECOVERED after=<n>s previously_fallen=<n>` · `COVERAGE-FALL-CHECK-ERROR detail="…"` (the child could not be read — and `unknown` is never written to state, so this line repeats every run and can never become a resting state that silences itself).
 - **The gate is the child's EXIT CODE, never its `ok` field** — the rule `[0.4.354]`/`[0.4.355]`/`[0.4.356]` established, applied a fourth time: rc **0/1** → a verdict, with `counts.unexplained == 0` → `ok`, a positive count → `fall` (the child's own finding, standing even when its JSON was unreadable — a fall that was seen is never turned into "could not verify"; the count is reported `?`), everything else (rc 3 refusal, another exit, timeout, missing binary, unreadable JSON) → `unknown`, never a pass and never a fall. M14 plants the defeated gate (rc 0 with `counts.unexplained=4` read as `ok`).
 - **Warn-only twice over, and the channel is split.** The child never changes the process exit code and never changes the monitor's `state`: an unexplained fall edits `coverage.alert` / the human line's `coverage_alert=`, while `alert=`/`state=` stay the monitor's own — a wrapper alert that could be mistaken for a tree verdict is exactly the confusion `[0.4.204]`'s channel rule exists to prevent. A monitor `CHECK-ERROR` plus a fallen child writes both lines, and the state file keeps the last SETTLED monitor verdict beside the child's own four fields.
-- **State, log line, JSON, human line.** `red-watch.state` gains `coverage` / `coverage_since` / `coverage_count` / `coverage_ids`; the dated log line gains its settled tail (`coverage=ok` / `fall:<n>` / `unknown`) after `baseline=`; the JSON payload gains the `coverage` object (`alert` / `count` / `ids` / `state`) — **16 → 17 keys**, suite F11 and the REGISTRY's key list updated in the same commit; the human line gains `coverage=… coverage_alert=…`. Two new env hooks, `REDWATCH_COVERAGE_GUARD_BIN` and `REDWATCH_COVERAGE_GUARD_TIMEOUT` (default **20**, validated positive-integer **before any I/O** beside the monitor's own timeout and the baseline-stale timeout — a typo runs no child and writes no log line); the child inherits this environment unchanged, so the app train's `GLADEX_COVERAGE_GUARD_*` overrides reach it without a second config.
+- **State, log line, JSON, human line.** `red-watch.state` gains `coverage` / `coverage_since` / `coverage_count` / `coverage_ids`; the dated log line gains its settled tail (`coverage=ok` / `fall:<n>` / `unknown`) after `baseline=`; the JSON payload gains the `coverage` object (`alert` / `count` / `detail` / `state`) — **16 → 17 keys**, suite F11 and the REGISTRY's key list updated in the same commit; the human line gains `coverage=… coverage_alert=…`. Two new env hooks, `REDWATCH_COVERAGE_GUARD_BIN` and `REDWATCH_COVERAGE_GUARD_TIMEOUT` (default **20**, validated positive-integer **before any I/O** beside the monitor's own timeout and the baseline-stale timeout — a typo runs no child and writes no log line); the child inherits this environment unchanged, so the app train's `GLADEX_COVERAGE_GUARD_*` overrides reach it without a second config.
 
 ### Tests
 - **`bash tests/test_red_watch.sh` → 329 passed / 0 failed, rc 0; `--mutations` → 394 passed / 0 failed, rc 0** (baseline at HEAD: **311 / 0**, all twelve pre-existing mutants applying — measured, not assumed). New **section J** (J1–J67) on a pinned static child (the LIVE coverage guard is never read, the suite's hermeticity contract): the `--format json` argv, the four `coverage*` state fields through onset → `SET` (with the original `since` kept) → `RECOVERED` (clearing them), dedupe across repeated identical readings, the rc-gate in both directions (rc 1 + garbage JSON is still `fall` with count `?`; rc 0 unreadable / contradicting / rc 7 / missing binary / timeout are `unknown` and never `ok`), the timeout bound and its pre-I/O validation, warn-only exits throughout, the monitor-`CHECK-ERROR`-plus-fallen-child channel split, the `GLADEX_*` passthrough, and the default `<repo>/tools/coverage-guard` path resolved against a sandbox fake repo (running the real child there would read the LIVE coverage). Four new mutants — **M13** the coverage alert suppressed, **M14** the rc-gate defeated, **M15** the dedupe defeated, **M16** the `since` on a SET restarted — each asserted to apply exactly once, to parse, and to redden THIS suite on its own intended assertion.
 - **Affected suites (measured this run)**: `test_registry_coverage.sh` → rc 0, **497 / 0**; `test_registry_figure_readers.sh` → **43 / 1** (single red **V14b** `shift-deploy` 132 vs 127 — the standing environmental red, measured identically by runs 959/962/963/964/965/966/967, not this run's); `test_monitor_cascade.sh` → **49 / 0**; `test_system_status_red_watch.sh` → **150 / 0**; `test_gladex_monitor.sh` → the failures exactly the run's own dirty tree plus the standing **A12/A30** (lena/mia/sofia failed units) and **A31/A32** (`src/php/download.php` drift) — no new fault class; `queue-source-check` → rc 0 on this entry's `### Queue` pointer; `repo-lint` → rc 0, **363** changelog headings with this entry (both re-read in the close).

> build · mimo-v2.6-flash-free
✱ Grep "coverage.*alert.*count|17 keys|COVERAGE-FALL" in repo/tools/REGISTRY.md · 7 matches
> build · mimo-v2.6-flash-free
→ Read repo/tools/REGISTRY.md [limit=30, offset=7800]
> build · mimo-v2.6-flash-free
$ sqlite3 /opt/startup/dev/data/messages.db "SELECT COUNT(*) FROM messages WHERE direction='investor_to_agent' AND read=0;" 2>/dev/null; sqlite3 /opt/startup/prod/data/messages.db "SELECT COUNT(*) FROM messages WHERE direction='investor_to_agent' AND read=0;" 2>/dev/null
0
0
$ cd /data/repo && ./tools/version-check; ./tools/budget-show | head -5; ./tools/inbox-status | tail -3; ls /data/repo/hiring/queue/*.json 2>/dev/null | wc -l; date -u +%H:%MZ
OK: dev=0.4.28 prod=0.4.28 (match)
=== Budget Status ===
Month:       2026-10
Allowance:   5.00 €
Spent:       0.00 € (no October charge in any file — MEASURED 2026-10-12: no public registry source can date a renewal (DENIC RDAP publishes only a "last changed" event, and does so for every .de domain; DENIC whois has no expiry field; the IANA RDAP bootstrap has no "de" entry), so the one open unknown is your billing cadence: monthly 1.50 as brief §4 states, or annual at the 2027-09-20 registration anniversary? Still NEEDS-INVESTOR, REPORT.md §14)
Remaining:   5.00 €
  INBOX.md        : 80 entries, 80 handled, 0 open

verdict: OK - nothing owed (0 unread, 0 open entries all replied)
2
13:14Z
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/coverage-guard --format json 2>/dev/null | python3 -c '
import json,sys
d=json.load(sys.stdin)
print("rc-field ok:",d.get("ok"),"counts:",d.get("counts"))
for r in d.get("rows",[]):
    if r.get("state")!="ok": print(" row:",r.get("identity"),r.get("state"),r.get("previous"),r.get("current"),r.get("unit_state"))
' ; echo "---- system-status row:"; ./tools/system-status 2>/dev/null | grep -i 'coverage\|baseline' 
Traceback (most recent call last):
  File "<string>", line 6, in <module>
AttributeError: 'str' object has no attribute 'get'
rc-field ok: False counts: {'ok': 3, 'explained': 0, 'unexplained': 4, 'no_record': 0, 'gone': 0}
---- system-status row:
  deploy-path               [OK] ledger OK - 48 file(s) across dev+prod attested (/data/agent-logs/promote-ledger.jsonl; source: baseline+promote)
  baseline-stale            [WARN] baseline stale: 3 transcript(s) above their recorded value - run tools/coverage-guard --record (green only)
  coverage-guard            [WARN] unexplained fall: 4 transcript(s) below their recorded value with a healthy unit (aylin,dispatcher,jonas,leon) - investigate with tools/coverage-guard (a real drop is accepted by editing agent-logs/coverage-baseline.json)
> build · mimo-v2.6-flash-free
$ cd /data/repo && tail -c 400 agent-logs/PROGRESS.md | cat -A | tail -3 | head -c 200; echo; echo "=== file end marker ==="
*no new fault class**; `./tools/queue-source-check` M-bM-^FM-^R **rc 0** on `[0.4.357]` pointer-only; `git status --porcelain` M-bM-^FM-^R **empty**; `git rev-list --count origin/main..main` M-bM-^FM-
=== file end marker ===
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/inbox-status
inbox-status — STEP 0 investor duty

  messages unread : dev 0 / prod 0 / total 0
  last reply      : dev 146 / prod 111
  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
$ cd /data/repo && cat >> agent-logs/PROGRESS.md <<'ENTRY'

## Builder — run 968 — **run 967's re-queued ALERTS half lands: `tools/red-watch` gains the coverage-guard reader, so the live `UNEXPLAINED FALL (aylin, dispatcher, jonas, leon)` reaches the `*/15` cron's own alert file** — four change-only alerts (onset / SET keeping `since` / RECOVERED / CHECK-ERROR repeating), rc-gated (never `ok`-gated), warn-only twice over (never the exit code, never the monitor's `state`) — suite **311 → 394** under `--mutations` (section J + M13–M16), no new test file (suite count stays **120**, `- Live:` untouched), `CHANGELOG.md` **[0.4.358]**; the five step files are still uncommitted at append time, so this run's own commit carries them **plus** this file (pathspec-staged, never `git add -A`), no `app/` file

- **STEP 0, paid first and re-read at the close — owed nothing, no row written.** `SELECT COUNT(*) … WHERE direction='investor_to_agent' AND read=0` → **0 dev / 0 prod** at orientation and again before this entry; `./tools/inbox-status` → **rc 0**, `OK - nothing owed (0 unread, 0 open entries all replied)`, last reply **dev 146 / prod 111**; `INBOX.md` → **80 entries / 80 handled / 0 open**. No `agent_to_investor` row inserted in either database, nothing marked read, `INBOX.md` untouched — an empty inbox is answered by the measurement, not by a status row. **Production is 0.4.28** (the investor's 2026-10-05 correction stands, the wrong figure quoted nowhere): `./tools/version-check` → `OK: dev=0.4.28 prod=0.4.28 (match)`; `./tools/budget-show` → **2026-10, allowance 5.00 / spent 0.00 / remaining 5.00**; `./tools/queue-source-check` before this entry → **rc 0**, `one queue: [0.4.357] pointer-only` (the naming AND the ordering counts both clear at this moment — `[0.4.358]` does not exist in the changelog yet at that reading, and this entry is what names it, so the R8 violation returns the instant this is appended and clears again with this run's own commit). `hiring/queue/` re-checked → **2 files, both still `status: "proposed"`** (`marco-steiner.json`, `ruben-stoll.json`) — items **(93)**/**(97)(a)** remain investor-gated. Spend **0.00**, free `*-free` models only, no key configured, **no secret or PII in any prompt, file or commit**.

- **Queue read, and the step taken: run 967's re-queued ALERTS half — and the first live firing landed BEFORE this run committed a byte.** The queue line, verbatim: *"the shape to price is the one run 965→966 took for the sibling: a `COVERAGE-FALL` reader beside `[0.4.356]`'s `BASELINE-STALE` reader, riding the existing `*/15` cron, change-only, rc-gated, warn-only twice over (never the exit code, never the monitor's `state`), at the same one-tool-one-suite price — a live-count threshold is NOT it"*. Alternatives measured out, not assumed: run 952's **(i)** future-dated-footer guard (*only on a measured future-dated marker* — still **0** found) and **(iii)** A12 wording (*only on a measured A12 misread* — the row already names mia + sofia and the `work/ownlog/other` split); **(93)**/**(97)(a)**, **(2'')**, **§14 renewal**, SOA MNAME, **(113)**'s `cp`s and **2026-10-20** §10 #6 each wait on investor/dated/operator gates; **A31/A32** wait on `b460155`'s `src/php/download.php` reaching an environment (the dev-sync desk's). Priced **one-tool-one-suite**, change-only, with the same two-part warn-only discipline the sibling reader already carries.

- **The change: one second reader inside `tools/red-watch`, gated on the child's EXIT CODE, never on its `ok` field.** The tool now runs `tools/coverage-guard --format json` once per tick beside its own monitor and beside `[0.4.356]`'s baseline reader, and follows the child's state with the same change-detection rule the monitor's verdict already uses. **The gate is the exit code, never `ok`**: the child prints `ok:false` for BOTH rc 1 (a real fall) and rc 3 (its refusal — no baseline, no coverage input), so rc **0/1** → a verdict with `counts.unexplained == 0` → `ok` and a positive count → `fall` (the child's own finding, standing even when its JSON was unreadable — a fall that was seen is never turned into "could not verify"; the count is then reported `?`), rc 3 / another exit / timeout / missing binary / unreadable JSON → `unknown`, **never a pass and never a fall**. M14 plants exactly the defeated gate (rc 0 with `counts.unexplained=4` read as `ok`). Four alerts, **change-detection only — never a threshold**: `COVERAGE-FALL since=<epoch> count=<n> identities="…" detail="…"` (onset, carrying `[0.4.357]`'s own finding wording so the alert greps identically to the dashboard row) · `COVERAGE-FALL-SET since=… count=… identities="…"` (the SET moved — the comparison is on WHICH identities are below their recorded value, not on how many, so a set that grows does not restart the window: the original `since` is kept, the same design as `RED-SET`) · `COVERAGE-FALL-RECOVERED after=<n>s previously_fallen=<n>` · `COVERAGE-FALL-CHECK-ERROR detail="…"` (and `unknown` is never written to state, so an unreadable child re-alerts every 15 minutes and can never become a resting state that silences itself). **Warn-only twice over, and the channel stays split**: the child changes neither the process exit code nor the monitor's `state` — an unexplained fall edits `coverage.alert` / the human line's `coverage_alert=`, while `alert=` / `state=` stay the monitor's own (the `[0.4.204]` channel rule), and the state file keeps the last SETTLED monitor verdict beside the child's four fields. `red-watch.state` gains `coverage` / `coverage_since` / `coverage_count` / `coverage_ids`; the dated log line gains its settled tail (`coverage=ok` / `fall:<n>` / `unknown`) after `baseline=`; the JSON payload gains the `coverage` object (`alert` / `count` / `detail` / `state`) — **16 → 17 keys**, suite F11 and the REGISTRY's key list moved in the same commit. Two env hooks, `REDWATCH_COVERAGE_GUARD_BIN` and `REDWATCH_COVERAGE_GUARD_TIMEOUT` (**20**, positive-integer validated **before any I/O**, beside the monitor's own timeout and the baseline reader's — a typo runs no child and writes no log line); the child inherits the environment unchanged, so the app train's `GLADEX_COVERAGE_GUARD_*` overrides reach it with no second config.

- **Suite **329 passed / 0 failed** (`bash tests/test_red_watch.sh`, rc 0); `--mutations` → **394 passed / 0 failed** (rc 0)**, measured on the detached-tree baseline of **311 / 0** (all twelve pre-existing mutants applying — measured, not assumed). New **section J** (J1–J67) on a pinned static child (the LIVE coverage guard is never read — the suite's hermeticity contract): the `--format json` argv · the four `coverage*` state fields through onset → `SET` (with the original `since` kept) → `RECOVERED` (clearing them) · dedupe across repeated identical readings · the rc-gate in both directions (rc 1 + garbage JSON is still `fall` with count `?`; rc 0 unreadable / contradicting / rc 7 / missing binary / timeout are `unknown` and never `ok`) · the timeout bound and its pre-I/O validation · warn-only exits throughout · the monitor-`CHECK-ERROR`-plus-fallen-child channel split · the `GLADEX_*` passthrough · and the default `<repo>/tools/coverage-guard` path resolved against a sandbox fake repo (running the real child there would read the LIVE coverage). Four new mutants — **M13** the coverage alert suppressed, **M14** the rc-gate defeated, **M15** the dedupe defeated, **M16** the `since` on a SET restarted — each asserted to apply exactly once, to parse, and to redden THIS suite on its own intended assertion (`M16 → the wrong verdict lands on the intended assertion` is the battery's own last line).

- **Mid-run events, each disclosed rather than absorbed:** (a) **the `*/15` cron runs the working tree, not the commit**, so the edited reader fired live at **12:30:01Z — the first live `COVERAGE-FALL` onset, ~45 minutes before this entry was written**: `2026-10-15T12:30:01Z COVERAGE-FALL since=1792067550 count=4 identities="aylin dispatcher jonas leon" detail="unexplained fall: 4 transcript(s) below their recorded value with a healthy unit (aylin,dispatcher,jonas,leon) - investigate with tools/coverage-guard (a real drop is accepted by editing agent-logs/coverage-baseline.json)"` — the reader's own onset, recorded by the very channel it was built for, with no commit involved. (b) **The 13:00:01Z tick measured the dedupe live**: no second `COVERAGE-FALL` line in `red-watch-ALERTS.log`, state file `coverage=fall / coverage_since=1792067550 / coverage_count=4 / coverage_ids=aylin dispatcher jonas leon` unchanged (`updated=2026-10-15T13:00:01Z`) — the "the SET did not change" branch firing on a standing fall, i.e. exactly the silence a threshold could not have produced. (c) The cross-channel agreement was **measured, not asserted**: at run 967's close the dashboard row read **3** (`aylin,jonas,leon`) and this run's alert reads **4** — re-reading both live now shows `system-status`'s `coverage-guard` row and `red-watch`'s state both at **4 (aylin,dispatcher,jonas,leon)**, i.e. the count moved 3 → 4 between the two runs (dispatcher's fall appeared) and the two channels agree on the same child's same reading; no discrepancy, and the reader fired on the change. (d) The first affected-suite batch ran 2-way parallel and its own 600 s runner timeout killed `test_registry_coverage.sh` once (**rc 124**, contention on this shared box) — re-run solo → **rc 0, 497 / 0**. (e) No concurrent identity touched this run's five step files during the window (`git status --porcelain` still shows exactly the five `M` lines this run wrote); nothing was swept and nothing was re-staged.

- **Verification, all on this tree, all taken before this entry.** Green — `red_watch` **329/0** and **394/0** under `--mutations`; `test_registry_coverage.sh` **497 / 0, rc 0** (C28g **32**, G38 **27** unchanged — no new `--` token pair, section F F3/F5/F6 green with **`figure_stale 0`** on the **120**-suite `- Live:` line); `test_monitor_cascade.sh` **49 / 0**; `test_system_status_red_watch.sh` **150 / 0**. Red, both standing — `test_registry_figure_readers.sh` **43 / 1**, single red **V14b `shift-deploy` `want=132 got=127`** (the standing environmental red, measured identically by runs 959/962/963/964/965/966/967, **not** this run's) and `test_gladex_monitor.sh` **28 assertions, 22 / 6** with the failures exactly **A3/A15** (this run's own dirty tree — the five uncommitted step files) + the standing **A12/A30** (`identity-run@lena/mia/sofia.service`, the failed set is now three units) and **A31/A32** (`src/php/download.php` drift) — **no new fault class**. `./tools/repo-lint` → **rc 0, 228 linted file(s) parse clean**, `362 changelog version heading(s)` **read off HEAD** — the working tree reads **363** (`grep -c '^## \['`), i.e. **362 → 363 with `[0.4.358]`, predicted and re-read after the commit rather than carried**. Live reads: `./tools/coverage-guard` → **exit 1, `counts {ok 3, explained 0, unexplained 4, no_record 0, gone 0}`**; `./tools/system-status` → both rows `[WARN]` (`baseline stale: 3 …` and `unexplained fall: 4 …`), human healthy line intact; `./tools/red-watch` human line → `coverage=fall coverage_alert=COVERAGE-FALL …` beside `baseline=stale`, monitor `state=`/`alert=` untouched; `--format json` → **17 keys** with the `coverage` object.

- **Scope and safety, one line each:** this run's own commit carries **five step files plus `agent-logs/PROGRESS.md`**, staged by **explicit pathspec** read from `git status --porcelain` immediately before staging — **never `git add -A`** on this shared index; **`rm` only inside `/tmp/opencode`** (the `run968/` logs); **no live messages-DB write**, no DNS write, no mail sent, no paid API, no API key configured; **`hiring/queue/` read-only** (2 proposals `status=proposed`, the investor's seat); **no `systemctl reset-failed`, no restart, no unit or timer change, no `identity-run.sh`/`sofia-watch` edit** — lena's, mia's and sofia's failed units left standing on purpose and read only; **the readers never write** — `coverage-guard --record` was NOT run, the live baseline and every transcript were READ ONLY, and this run edited no `agent-logs/coverage-baseline.json`; **`tests/test_registry_coverage.sh` NOT edited this run** (C28g 32, G38 27) and no existing REGISTRY heading renamed or moved; **never `git stash` / `git stash pop`** in this tree (run 959's measured loss); **no `app/` file edited** → no reviewer gate, nothing to promote, dev and prod untouched at **0.4.28**; spend **0.00**, free `*-free` models only.

- **Queue — next small step (read this first):** **STEP 0 first every run** (today **0 unread dev / 0 prod**, `INBOX.md` **80/80/0 open** — an empty inbox is answered by the measurement). **This run closed run 967's re-queued ALERTS half** (`COVERAGE-FALL` reader in `tools/red-watch`, sections J1–J67 + M13–M16). **NEW, re-queued unnumbered: the live-verification half of the same finding, and only on a measurement** — both readers now exist on both channels, so the remaining gap is that **nothing asserts the two channels AGREE** (this run had to measure it by hand: `system-status` row 4 vs `red-watch` state 4, dispatcher having joined the fall set between two runs). The shape to price is a cross-reader assertion, not a new tool: a suite that runs `tools/coverage-guard --format json` **once** and asserts the `system-status` row and a `red-watch`-shaped verdict derive the SAME count and the SAME identity list from that one reading — i.e. the readers may not each re-derive the set. **Taken only if the next measurement still exposes a disagreement or a re-derivation** (both readers read the child's `rows` today, so the honest check is whether that is true in code and not just in prose); if it holds, this line closes itself and the next measurement owns the next step. **Also still open:** run 952's **(i)** future-dated-footer guard *only on a measured future-dated marker* (still **0** found) and **(iii)** A12 wording *only on a measured A12 misread*; **(93)**/**(97)(a)** first if the investor answers `hiring/queue/*.json` (**2 proposals**, unchanged since 2026-10-04 / 2026-09-30); **(2'')** the day the `GETTING_STARTED.md` keep/fold/retire answer arrives; **§14 renewal** cadence NEEDS-INVESTOR; the LAN-only A/B/C; NUL-blob watch on a measured harm only; **(av)(d)** after the `REPORT.md` §14 sign-off; **(i)** until the prune lands; SOA MNAME and Pip's mail-policy research NEEDS-INVESTOR; **(113)**'s two `cp`s operator-blocked; **2026-10-20** §10 #6 re-read; the `/stats` display row; and **(ax)**, **(e)**, the drift-verify line, **(aw)**, the `promote-gates` verdict watch, and run 915's monitor item — **A31/A32 now waiting on `b460155`'s `src/php/download.php` reaching dev and prod** (the dev-sync desk's, from `7f3166e`); **the live fall set (aylin, dispatcher, jonas, leon) and the stale set (lena, mia, sofia) are other seats' to fix** — the baseline only moves through `coverage-guard --record` green-only or a hand edit, and this seat runs neither.

- **Close-state:** this entry is appended **before** the commit that carries it (CLOSE PROTOCOL rule 1) and is never `git checkout --`-ed, staged-only or parked in `/tmp` (rules 2–3); the stop-state check is `git status --porcelain` **not listing `agent-logs/PROGRESS.md`** after the commit. The push must leave `git rev-list --count origin/main..main` → **0**. Post-push re-reads (`queue-source-check` → rc 0 on `[0.4.358]` pointer-only, `repo-lint` → **363** headings / 228 linted files / 0 missing, `red_watch` **329/0** and **394/0**, regcov **497/0**, `registry_figure_readers` (V14b standing), `system_status_red_watch` **150/0**, the monitor (expect A3/A15 cleared by the push, standing A12/A30/A31/A32), `inbox-status`, `version-check`, `budget-show`, live `red-watch --format json` with 17 keys, `git status --porcelain`, `rev-list`) plus the next `*/15` red-watch tick after the push are recorded in the **close-correction** bullet appended before any close commit (rule 2).
ENTRY
echo "appended"; tail -3 agent-logs/PROGRESS.md | head -1 | cut -c1-120
appended
- **Queue — next small step (read this first):** **STEP 0 first every run** (today **0 unread dev / 0 prod**, `INBOX.m
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/queue-source-check; echo "qsc rc=$?"; git status --porcelain
queue-source-check: OK - one queue: [0.4.357] pointer-only, 111 item line(s) frozen across 312 section(s), 203 PROGRESS bullet(s), 2 path token(s)
qsc rc=0
 M CHANGELOG.md
 M agent-logs/PROGRESS.md
 M tests/test_red_watch.sh
 M tools/REGISTRY.md
 M tools/red-watch
> build · mimo-v2.6-flash-free
$ cd /data/repo && git add tools/red-watch tests/test_red_watch.sh tools/REGISTRY.md CHANGELOG.md agent-logs/PROGRESS.md && git diff --cached --name-only && git commit -q -m "run 968 close: COVERAGE-FALL reader in red-watch ([0.4.358]) - section J1-J67 + M13-M16 (329/0, 394/0), REGISTRY + CHANGELOG, PROGRESS entry; first live onset fired 12:30:01Z count=4" && git log --oneline -1 && git status --porcelain; echo "porcelain-empty-rc=$?"; git rev-list --count origin/main..HEAD
CHANGELOG.md
agent-logs/PROGRESS.md
tests/test_red_watch.sh
tools/REGISTRY.md
tools/red-watch
13ae819 run 968 close: COVERAGE-FALL reader in red-watch ([0.4.358]) - section J1-J67 + M13-M16 (329/0, 394/0), REGISTRY + CHANGELOG, PROGRESS entry; first live onset fired 12:30:01Z count=4
porcelain-empty-rc=0
1
> build · mimo-v2.6-flash-free
$ cd /data/repo && git push origin main 2>&1 | tail -3; echo "push rc=$?"; git rev-list --count origin/main..main
To git://git.gladex.de/gladex.git
   7a0f509..13ae819  main -> main
push rc=0
0
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/queue-source-check; echo "qsc rc=$?"; ./tools/repo-lint 2>&1 | tail -4; echo "lint rc=${PIPESTATUS[0]}"; grep -c '^## \[' CHANGELOG.md; ./tools/version-check; ./tools/inbox-status | tail -2; ./tools/budget-show | sed -n '2,6p'
queue-source-check: OK - one queue: [0.4.357] pointer-only, 111 item line(s) frozen across 312 section(s), 203 PROGRESS bullet(s), 2 path token(s)
qsc rc=0
  go-compile: 47 module file(s) compile clean (1.979s)
  changelog-version: 363 changelog version heading(s), 363 unique, 10848 citation(s) checked, 0 missing, 13484 bare token(s) counted
  elf-allowlist: 2 tracked ELF file(s), all shipped
result: all 228 linted file(s) parse clean (exit 0)
lint rc=0
363
OK: dev=0.4.28 prod=0.4.28 (match)

verdict: OK - nothing owed (0 unread, 0 open entries all replied)
Month:       2026-10
Allowance:   5.00 €
Spent:       0.00 € (no October charge in any file — MEASURED 2026-10-12: no public registry source can date a renewal (DENIC RDAP publishes only a "last changed" event, and does so for every .de domain; DENIC whois has no expiry field; the IANA RDAP bootstrap has no "de" entry), so the one open unknown is your billing cadence: monthly 1.50 as brief §4 states, or annual at the 2027-09-20 registration anniversary? Still NEEDS-INVESTOR, REPORT.md §14)
Remaining:   5.00 €

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

[Thu Oct 15 15:15:21 2026] 127.0.0.1:43712 Accepted
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43712 Closing
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43728 Accepted
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43728 Closing
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43742 Accepted
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43742 Closing
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43756 Accepted
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43756 Closing
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43758 Accepted
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43758 Closing
[Thu Oct 15 15:15:21 2026] 127.0.0.1:43774 Accepted
[Thu Oct 15 15:15:22 2026] 127.0.0.1:43774 Closing
[Thu Oct 15 15:15:22 2026] 127.0.0.1:43778 Accepted
[Thu Oct 15 15:15:22 2026] 127.0.0.1:43778 Closing
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37070 Accepted
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37070 Closing
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37086 Accepted
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37086 Closing
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37102 Accepted
[Thu Oct 15 15:15:24 2026] 127.0.0.1:37102 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37112 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37112 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37118 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37118 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37126 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37126 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37130 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37130 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37144 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37144 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37152 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37152 Closing
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37164 Accepted
[Thu Oct 15 15:15:25 2026] 127.0.0.1:37164 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48580 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48580 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48584 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48584 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48598 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48598 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48608 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48608 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48614 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48614 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48620 Accepted
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48620 Closing
[Thu Oct 15 15:16:34 2026] 127.0.0.1:48634 Accepted
[Thu Oct 15 15:16:35 2026] 127.0.0.1:48634 Closing
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41240 Accepted
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41240 Closing
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41244 Accepted
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41244 Closing
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41246 Accepted
[Thu Oct 15 15:16:35 2026] 127.0.0.1:41246 Closing
[Thu Oct 15 15:20:24 2026] 127.0.0.1:60164 Accepted
[Thu Oct 15 15:20:24 2026] 127.0.0.1:60164 Closing
[Thu Oct 15 15:21:59 2026] 127.0.0.1:57940 Accepted
[Thu Oct 15 15:21:59 2026] 127.0.0.1:57940 Closing
[Thu Oct 15 15:21:59 2026] 127.0.0.1:57956 Accepted

Other app logs: downloads.log