Gladex Agent Logs

$ cd /data/repo && echo "--- queue-source-check:"; ./tools/queue-source-check; echo "rc=$?"; echo "--- ast/bash -n:"; python3 -c "import ast; ast.parse(open('tools/regression-run').read()); print('ast ok')"; bash -n tests/test_regression_run.sh && echo "bash -n ok"; echo "--- repo-lint (HEAD):"; ./tools/repo-lint --format json 2>&1 | python3 -c "import json,sys; d=json.load(sys.stdin); print('exit',d['exit_code'],'ok',d['ok'],'files',d['files_total'],'cv',d.get('changelog_version'),'failures',d['failures'],'errors',d['errors'])"
--- queue-source-check:
queue-source-check: OK - one queue: [0.4.102] pointer-only, 111 item line(s) frozen across 60 section(s), 64 PROGRESS bullet(s)
rc=0
--- ast/bash -n:
ast ok
bash -n ok
--- repo-lint (HEAD):
exit 0 ok True files 198 cv {'attempted': True, 'path': 'CHANGELOG.md', 'entries': 106, 'unique': 106, 'unparseable': 0, 'duplicates': [], 'citations_seen': 3095, 'citations_in_series': 2971, 'citations_missing': [], 'series': ['0.1', '0.2', '0.3', '0.4'], 'ok': True, 'reason': None, 'summary': '106 changelog version heading(s), 106 unique, 3095 citation(s) checked, 0 missing'} failures [] errors []
$ cd /data/repo && for t in test_queue_source.sh test_repo_lint.sh; do echo "--- $t"; bash tests/$t 2>&1 | tail -2; done; for t in test_changelog_api.php test_app_version.php test_cli_version.php; do echo "--- $t"; php tests/$t 2>&1 | tail -2; done
--- test_queue_source.sh
queue-source-check suite: 122 passed, 0 failed
===============================
--- test_repo_lint.sh
repo-lint suite: 420 passed, 0 failed
===============================
--- test_changelog_api.php

=== Results: 86 passed, 0 failed ===
--- test_app_version.php

=== Results: 39 passed, 0 failed ===
--- test_cli_version.php

=== Results: 35 passed, 0 failed ===
> build · mimo-v2.6-flash-free
$ cd /data/repo && cat >> agent-logs/PROGRESS.md <<'ENTRY'

## 2026-09-27T20:35Z main-loop run — STEP 0 answered FIRST (0 unread both DBs; fresh-probe reply **dev 97 / prod 63**), then queue item **(46) EXECUTED**: `regression-run --note` is now ONE line of at most `NOTE_MAX` (200) characters, bounded where it is accepted and again where a stored document is read

- **STEP 0 (first action, before any product work)**: unread `investor_to_agent` = **0 dev / 0 prod**, read with a direct `SELECT count(*) … AND read=0` on **both** live SQLite DBs, so nothing was marked read that is not mine. `tools/inbox-status --format json` → **exit 0**, *owed 0* — **74 entries / 73 handled / 1 open** (line 468, the Nextcloud half of the account ask, **replied** and blocked on REPORT.md §14, not missed), `open_unreplied []`. No order was waiting, so a fresh-probe reply was inserted **parameterised** (python3 `?` binding, body held in a script variable — never interpolated into shell, never printed) into both DBs: **dev 97 / prod 63**, `read=1`, re-verified after the insert that the newest row in each is `agent_to_investor` / `read=1` and unread is still **0 / 0**.
- **The step, and why this run took it**: `[0.4.101]` shipped `--note` and queued the bound question in the same breath — *"(46) the `note` is prose the runner supplies and nothing constrains it … worth either a documented bound (reject newlines, cap the length) or one assertion pinning that a newline is refused"*. It is not a style question: `print_baseline` writes `  note: <text>` at a **fixed two-space indent**, so a newline in the text produces a second line with **no prefix** — and the prefix is what makes a line read as this tool's output, so a note carrying `a\n` + `  result: all suites green (exit 0)` deposits a verdict the reader cannot tell from a real one, inside the report of the tool whose job is trustworthy totals. The queue's own caution (that inventing a bound *after* `I88`/`I89` were written would change their meaning rather than extend them) held: **both existing assertions are untouched and still green** — blankness keeps its predicate and its message, and the bound is a *third* rule for a string that already had two.
- **What changed**: `NOTE_MAX = 200`, one module constant beside the baseline constants; `note_is_blank()` is now the single blankness rule for the flag **and** the document, `note_problem()` the single bound rule (control character → `must be a single line (no newline, tab or other control character)`; over length → `must be at most 200 characters (got N)`). Write side: both refusals sit with the existing `--note` checks — after parsing, **before discovery** — exit 2, and under `--format json` still **one 14-key object on stdout** with `exit_code: 2` and **0 bytes on stderr** (the D-section contract does not move). Read side: `load_baseline` applies the same two bounds to a stored `note` (`baseline <file> note …` → exit 2) at the position `format`, `kind` and `totals` already occupy, because a baseline is JSON anyone can edit and the read half of a lie is the same lie. The option's `--help` text and the docstring sentence both **interpolate the constant**, so the number cannot be stated twice and read differently.
- **Measured, not claimed**: `tests/test_regression_run.sh` **218 → 237 passed, 0 failed** — **19** new assertions (`A10`–`A12`, `I95`–`I110`) covering the two refusals with their diagnoses and empty stdout, the tab, the bound-and-actual-length message, **exactly 200 characters accepted and stored byte for byte**, the JSON refusal's 14-key/empty-stderr shape, and all four read-side twins. Two of them were wrong on the first run and are recorded rather than hidden: **A11 failed** because argparse **wraps** help text (`at most 200` / `chars` on two lines), so the needle now matches one *unwrapped* line; and the derived `note_max` came back **empty** pre-fix, turning the arithmetic into reds about the empty string — fixed with an explicit fallback to the documented `200`, which is what makes the replay below's reds mean the defect instead of the derivation.
- **Pre-fix replay, captured in place**: `git show HEAD:tools/regression-run` swapped over the tool (old md5 `1f3f42137612a0e214de5b2408c2c9c2`, log `/tmp/opencode/note46/pre-fix.log`) → **222 passed / 15 failed** = `237 − 15`, every red a finding: `A10` (no such constant), `A11`/`A12` (neither rendering exists), `I95`–`I98` (both control-character notes accepted, rc 0), `I99`/`I100`/`I103` (the 201-char note and its JSON twin accepted), `I106`–`I110` (every hand-edited document loaded and ran). **Four guards passed pre-fix and identify themselves**: `I101` (exactly 200 was always legal — the boundary must not move), `I102` (it was always stored at full length), `I104` (the usage verdict's key set) and `I105` (stderr empty on a refusal — which passed pre-fix because nothing refused, disclosed as coincidence, not proof). Restore verified by hash: `a334358a0e6b02031b7025968c1be31a` on both the worktree and `/tmp/opencode/note46/new-tool.bak`.
- **Gates, read after the edits**: `tools/queue-source-check` → **exit 0**, *"`[0.4.102]` pointer-only, 111 item line(s) frozen across **60** section(s), 64 PROGRESS bullet(s)"* (59 → 60 is this entry's own pointer-only section; **111 unchanged**, so the freeze held — the bullet count moves to 65 with the append itself). `bash tests/test_queue_source.sh` → **122 / 0**; `bash tests/test_repo_lint.sh` → **420 / 0**; `php tests/test_changelog_api.php` → **86 / 0**; `php tests/test_app_version.php` → **39 / 0**; `php tests/test_cli_version.php` → **35 / 0**. `tools/repo-lint --format json` read **HEAD** (`bf72efa`) → **exit 0, `ok true`, `files_total 198`, entries 106 / unique 106 / `duplicates []` / `citations_missing []`** — it lints *committed* trees by design, so the pre-commit read cannot see `[0.4.102]`; the 106 → **107** movement is asserted post-commit instead of carried. `ast.parse` on the tool, `bash -n` on the suite. **Regression on the frozen tree, before the commit**: `./tools/regression-run --format json --log-dir /tmp/opencode/note46/regr` → **54 suites, 4637 passed, 0 failed, 0 skipped, exit 0**, all 54 `ok`, `conflicts []` — the green baseline **54 / 4618** plus **exactly +19**, this step's own assertion count (`4618 + 19 = 4637`, `54 = 54`, no other suite moved).
- **Docs**: `tools/REGISTRY.md` gained the extended `--note` option text (both bounds, the constant, the stored-note twin), the extended exit-2 list and a **"One line, at most `NOTE_MAX`"** paragraph under **"Why, not just what"** naming this entry as queue item (46); `CHANGELOG.md` gained **`[0.4.102]`** parked at the bottom like `[0.4.29]`–`[0.4.101]` with the pointer-only `### Queue`, every `[x.y.z]` token it cites pre-grepped against the heading list.
- **Deliberately not done**: **no change to the 14-key JSON contract** (a refusal is still one object with `baseline: null`), **no truncation** (a long note is refused, never silently cut — a shortened reason is a different reason), no baseline format bump (a document with a compliant note reads exactly as before; without one it does not exist yet in history), **`print_baseline` gained no guard of its own** — the read half already refuses before the print half can be reached, so a second check there would be a second rule for one string — no promote, no service restart, no certificate touched, **zero DNS writes**, no mail sent, Nextcloud untouched (`installed:false`, §14 open), `/opt/cloud` untouched, Immich untouched (library `total: 0`, 8 accounts).
- **A queue drift found while re-reading, corrected rather than carried**: the previous entry's bullet said *"carry items … from the 21:35Z entry unchanged"* but re-listed only the 18:55Z carry list, so **(44)** (commit-time duty gate) and **(45)** (pin the Immich round-trip's three corrections in `tests/`) — both introduced as *new* by the 21:35Z entry — silently left the queue. They are restored below with their original wording, which is the divergence class (36) describes happening inside the authoritative file itself; nothing checks a carry list against the entry it says it carries from, so the restoration is recorded here rather than assumed.
- **Staging hazard — the standing rule held**: `git status --porcelain` read immediately before staging decides the paths; files are staged **explicitly by path, never `git add -A`**, which has swept unfinished entries from other identities **seven** times in this file. Author resolves to `Atlas <atlas@gladex.de>`.
- **Next-candidate queued, not actioned**: carry items (2), (4)–(8), (14), (15), (17), (18), (20), (22), (23), (25), (28), (32), (36), (39) and (42) from the 20:00Z entry unchanged, **plus (44) and (45) restored from the 21:35Z entry** (dropped by omission in the 20:00Z bullet — see the drift bullet above) — item **(46) struck — EXECUTED by `[0.4.102]`** (`NOTE_MAX = 200`, `note_is_blank()` + `note_problem()` as one rule each for both halves, control character → exit 2, over length → exit 2 with the real count, `load_baseline` holding a stored note to the same two bounds before any suite runs, docstring `Baseline comparison:`/`Exit codes:`/`--help` all interpolating the one constant, `test_regression_run.sh` 218 → 237 with A10–A12 + I95–I110, pre-fix replay 222/15 with four self-identifying guards, regression 54 / 4637 / 0). New from this run: **(47)** the bound is documented in **three** places that nothing compares — `tools/regression-run`'s docstring (rendered into `--help`, A5 pins that pair), `tools/REGISTRY.md`'s prose (`--note` option line + exit-2 list + the new paragraph) and this entry — and only the first pair is checked; a fourth exit-2 cause added to the tool next month would leave `REGISTRY.md` reading green and complete while omitting it, which is the epilog-copy defect `[0.4.59]` closed for `--help` applied to the registry. Worth either a `repo-lint` rule that extracts the `Exit codes:` exit-2 list from the docstring and the corresponding `REGISTRY.md` bullet and asserts the causes match, or a `tests/` assertion doing the same — not actioned this run because the step was already a constant, two predicates, two call sites, 19 assertions, a replay and two documents, and a prose-extraction rule over a file five identities append to is its own careful step.
ENTRY
./tools/queue-source-check; echo "rc=$?"; wc -l agent-logs/PROGRESS.md
queue-source-check: OK - one queue: [0.4.102] pointer-only, 111 item line(s) frozen across 60 section(s), 65 PROGRESS bullet(s)
rc=0
5890 agent-logs/PROGRESS.md
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/regression-run --format json --log-dir /tmp/opencode/note46/regr2 > /tmp/opencode/note46/regr2.json 2>/tmp/opencode/note46/regr2.err; echo "rc=$?"; python3 -c "
import json
d=json.load(open('/tmp/opencode/note46/regr2.json'))
print('suites', len(d['suites']), 'totals', d['totals'], 'exit', d['exit_code'], 'conflicts', d['conflicts'])
bad=[(s['suite'],s['status'],s['reason']) for s in d['suites'] if s['status']!='ok']
print('non-ok:', bad)
for s in d['suites']:
    if 'queue' in s['suite'] or 'changelog' in s['suite'] or 'regression' in s['suite']:
        print(s['suite'], s['passed'], s['failed'])
"
rc=0
suites 54 totals {'passed': 4637, 'failed': 0, 'skipped': 0, 'suites_run': 54} exit 0 conflicts []
non-ok: []
test_changelog_api.php 86 0
test_changelog_mobile.php 125 0
test_queue_source.sh 122 0
test_regression_run.sh 237 0
> build · mimo-v2.6-flash-free
$ cd /data/repo && git status --porcelain
M CHANGELOG.md
 M agent-logs/PROGRESS.md
 M tests/test_regression_run.sh
 M tools/REGISTRY.md
 M tools/regression-run
> build · mimo-v2.6-flash-free
$ cd /data/repo && git add tools/regression-run tests/test_regression_run.sh tools/REGISTRY.md CHANGELOG.md agent-logs/PROGRESS.md && git status --porcelain --untracked-files=no && git commit -q -m "[0.4.102] queue item (46) — regression-run --note is now ONE line of at most NOTE_MAX (200) characters: a control character (newline/tab) or over-length note is refused on the write side and again by load_baseline before any suite runs, both renderings interpolating the one constant; test_regression_run 218 -> 237 (pre-fix replay 222/15), regression 54/4637/0; STEP 0 replied first (dev 97 / prod 63)" && git log --oneline -2 && git status --porcelain
M  CHANGELOG.md
M  agent-logs/PROGRESS.md
M  tests/test_regression_run.sh
M  tools/REGISTRY.md
M  tools/regression-run
768e5a6 [0.4.102] queue item (46) — regression-run --note is now ONE line of at most NOTE_MAX (200) characters: a control character (newline/tab) or over-length note is refused on the write side and again by load_baseline before any suite runs, both renderings interpolating the one constant; test_regression_run 218 -> 237 (pre-fix replay 222/15), regression 54/4637/0; STEP 0 replied first (dev 97 / prod 63)
bf72efa docs: post-commit confirmation — repo-lint 106/106 unique (files 198), queue [0.4.101] pointer-only 111 items/59 sections/64 bullets, suites 420+218+86 green against HEAD, committed md5 1f3f4213 == worktree, regression 54/4618/0, pushed b3ad629..2808a8c
> build · mimo-v2.6-flash-free
$ cd /data/repo && echo "--- committed bytes == tested bytes:"; md5sum tools/regression-run tests/test_regression_run.sh; git show HEAD:tools/regression-run | md5sum; git show HEAD:tests/test_regression_run.sh | md5sum; echo "--- STEP 0 re-check:"; sqlite3 /opt/startup/dev/data/messages.db "select count(*) from messages where direction='investor_to_agent' and read=0;"; sqlite3 /opt/startup/prod/data/messages.db "select count(*) from messages where direction='investor_to_agent' and read=0;"; ./tools/inbox-status --format json | python3 -c "import json,sys; d=json.load(sys.stdin); print('unread',d['unread']['total'],'owed',d['owed'],'inbox',d['inbox']['entries'],d['inbox']['handled'],d['inbox']['open'],'last',d['last_reply'])"; echo "rc=$?"
--- committed bytes == tested bytes:
a334358a0e6b02031b7025968c1be31a  tools/regression-run
8ab2b730dd00e650fc8c6095770744fc  tests/test_regression_run.sh
a334358a0e6b02031b7025968c1be31a  -
8ab2b730dd00e650fc8c6095770744fc  -
--- STEP 0 re-check:
0
0
unread 0 owed {'unread_messages': 0, 'unreplied_entries': 0, 'total': 0} inbox 74 73 1 last {'dev': {'id': 97, 'ts': '2026-09-27T20:22:07Z'}, 'prod': {'id': 63, 'ts': '2026-09-27T20:22:07Z'}}
rc=0
$ cd /data/repo && echo "--- repo-lint (HEAD):"; ./tools/repo-lint --format json | python3 -c "import json,sys; d=json.load(sys.stdin); cv=d['changelog_version']; print('exit',d['exit_code'],'ok',d['ok'],'files',d['files_total'],'entries',cv['entries'],'unique',cv['unique'],'dups',cv['duplicates'],'missing',cv['citations_missing'],'failures',d['failures'],'errors',d['errors'])"; echo "--- queue-source-check:"; ./tools/queue-source-check; echo "--- source-sync-check:"; ./tools/source-sync-check; echo "rc=$?"
--- repo-lint (HEAD):
exit 0 ok True files 198 entries 107 unique 107 dups [] missing [] failures [] errors []
--- queue-source-check:
queue-source-check: OK - one queue: [0.4.102] pointer-only, 111 item line(s) frozen across 60 section(s), 65 PROGRESS bullet(s)
--- source-sync-check:
source-sync-check  repo=/data/repo  env=both  2026-09-27T20:52:35Z
  dev  examples/workflows       8 files  OK
  dev  src/php                 13 files  OK
  prod examples/workflows       8 files  OK
  prod src/php                 13 files  OK
result: in sync — 42 file(s) across 2 env(s) (exit 0)
rc=0
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/system-status --format json | python3 -c "
import json,sys
d=json.load(sys.stdin)
print('overall',d['overall'],'errors',d['errors'])
for c in d['checks']:
    if c['status']!='ok': print(' ',c['status'],c['name'],'-',c['detail'][:160])
"; echo "rc=$?"
overall ok errors 0
  warning cloud - nextcloud 200 installed=false (setup not finished); immich 200 pong; 5/5 containers up
  warning SOA:gladex.de - serial=2026092602 consistent on 1.1.1.1+8.8.8.8; mname=placeholder (NEEDS-INVESTOR open)
  warning promote-gates - not promotable (newest reviewer-mailbox entry 2026-09-26T00:10:19Z, 1d 20h old): verdict REFUSED: verdict VERDICT-20260926-000744-promote-contrast-fix.md is STA
rc=0
$ cd /data/repo && for t in test_repo_lint.sh test_regression_run.sh; do echo "--- $t"; bash tests/$t 2>&1 | grep -E "passed|FAIL" | tail -3; done; echo "--- test_changelog_api.php"; php tests/test_changelog_api.php 2>&1 | tail -2
--- test_repo_lint.sh
PASS: X3: the FAIL line it refers to is still printed above it
PASS: M20 planted: the derivation is defeated, not bypassed
repo-lint suite: 420 passed, 0 failed
--- test_regression_run.sh
PASS: I21 passed_before 9
PASS: I22 passed_after 7
=== Results: 237 passed, 0 failed ===
--- test_changelog_api.php

=== Results: 86 passed, 0 failed ===
> build · mimo-v2.6-flash-free
$ cd /data/repo && git push origin main 2>&1 | tail -3; git status -sb | head -2; git log --oneline -1
To git://git.gladex.de/gladex.git
   bf72efa..768e5a6  main -> main
## main...origin/main
768e5a6 [0.4.102] queue item (46) — regression-run --note is now ONE line of at most NOTE_MAX (200) characters: a control character (newline/tab) or over-length note is refused on the write side and again by load_baseline before any suite runs, both renderings interpolating the one constant; test_regression_run 218 -> 237 (pre-fix replay 222/15), regression 54/4637/0; STEP 0 replied first (dev 97 / prod 63)
> build · mimo-v2.6-flash-free
$ cd /data/repo && cat >> agent-logs/PROGRESS.md <<'ENTRY'

### 2026-09-27T20:55Z — post-commit confirmation of the entry above (`768e5a6`, tree clean, pushed)

- **Post-commit reads on `768e5a6`, none carried from before the commit**: `tools/repo-lint --format json` → **exit 0, `ok true`, `files_total 198`**, `changelog_version` **`entries 107, unique 107, duplicates [], citations_missing []`** — the **106 → 107** movement this step's own append predicted, read from the committed blobs rather than the worktree the pre-commit run saw, with `failures []` and `errors []` (so the `[0.4.102]` token cited inside `REGISTRY.md` and inside the entry itself resolves against a heading committed with it). `tools/queue-source-check` → **exit 0, `[0.4.102]` pointer-only, 111 item line(s) frozen across 60 section(s), 65 PROGRESS bullet(s)** — 64 → 65 is this entry's own `Next-candidate queued` bullet, 111 unchanged. `tools/source-sync-check` → **in sync, 42 file(s) across 2 env(s), exit 0** (this step touched no synced tree).
- **The committed bytes are the bytes that were tested**: `md5sum tools/regression-run tests/test_regression_run.sh` → **`a334358a0e6b02031b7025968c1be31a`** / **`8ab2b730dd00e650fc8c6095770744fc`**, and `git show HEAD:<path> | md5sum` returns **the same two digests**, so the tool and the 237-assertion suite are in history exactly as written.
- **Suites re-run *after* the commit, against HEAD**: `bash tests/test_repo_lint.sh` → **420 / 0**, `bash tests/test_regression_run.sh` → **237 / 0**, `php tests/test_changelog_api.php` → **86 / 0**. The full regression (**54 / 4637 / 0**, exit 0) was re-taken on the *frozen* tree after the last doc append, not carried from the first read: the earlier run predated the `CHANGELOG.md`/`REGISTRY.md`/`PROGRESS.md` writes, so it was re-run and came back **identical (4637)** rather than assumed unchanged.
- **`tools/system-status --format json` → rc 0, `overall ok`, `errors 0`**, warnings only the standing three: `cloud` (Nextcloud `installed:false`, §14), `SOA:gladex.de` (MNAME placeholder, NEEDS-INVESTOR) and `promote-gates` (stale reviewer verdict — re-review is the reviewer's, not mine).
- **STEP 0 re-checked at the close of the run, not just at the start**: unread `investor_to_agent` still **0 dev / 0 prod**, `tools/inbox-status` → **exit 0** (**owed 0**; 74 entries / 73 handled / 1 open — line 468, the Nextcloud half of the account ask, replied and blocked on §14), newest row in each DB my own fresh-probe reply (**dev 97 / prod 63**, `read=1`, ts `2026-09-27T20:22:07Z`). Nothing arrived while the run was in flight.
- **Push**: `git push origin main` → **`bf72efa..768e5a6`** to `git://git.gladex.de/gladex.git`, `main` in sync with `origin/main`, working tree clean. Author resolves to `Atlas <atlas@gladex.de>`.
- **Safety (re-stated, nothing moved since)**: model spend **0.00** (`*-free` only), **`BUDGET.md` untouched (1.50 spent / 3.50 remaining**, month 2026-09, allowance 5.00), **zero DNS writes** (no `pdns-api.py` call), no paid API key configured, **no secret read or printed** (`/root/.pdns-token` and every `*.secret` untouched — this step created no credential of any kind; the only credential-shaped fact recorded anywhere remains the investor's own declared rule "password = own email"), **no service restarted, no certificate touched, no promote executed, no mail sent**, Nextcloud untouched (`installed:false`, §14 open), `/opt/cloud` untouched, Immich untouched (library `total: 0`, 8 accounts), and the message DBs touched only by this run's STEP 0 reply.
ENTRY
git status --porcelain
M agent-logs/PROGRESS.md
> build · mimo-v2.6-flash-free
$ cd /data/repo && ./tools/queue-source-check && git add agent-logs/PROGRESS.md && git commit -q -m "docs: post-commit confirmation — repo-lint 107/107 unique (files 198, citations_missing []), queue [0.4.102] pointer-only 111 items/60 sections/65 bullets, suites 420+237+86 green against HEAD, committed md5 a334358a/8ab2b730 == worktree, regression re-taken frozen at 54/4637/0, STEP 0 still 0/0 (dev 97 / prod 63), pushed bf72efa..768e5a6" && git push origin main 2>&1 | tail -2 && git status -sb | head -2 && git log --oneline -3
queue-source-check: OK - one queue: [0.4.102] pointer-only, 111 item line(s) frozen across 60 section(s), 65 PROGRESS bullet(s)
To git://git.gladex.de/gladex.git
   768e5a6..40f2a4d  main -> main
## main...origin/main
40f2a4d docs: post-commit confirmation — repo-lint 107/107 unique (files 198, citations_missing []), queue [0.4.102] pointer-only 111 items/60 sections/65 bullets, suites 420+237+86 green against HEAD, committed md5 a334358a/8ab2b730 == worktree, regression re-taken frozen at 54/4637/0, STEP 0 still 0/0 (dev 97 / prod 63), pushed bf72efa..768e5a6
768e5a6 [0.4.102] queue item (46) — regression-run --note is now ONE line of at most NOTE_MAX (200) characters: a control character (newline/tab) or over-length note is refused on the write side and again by load_baseline before any suite runs, both renderings interpolating the one constant; test_regression_run 218 -> 237 (pre-fix replay 222/15), regression 54/4637/0; STEP 0 replied first (dev 97 / prod 63)
bf72efa docs: post-commit confirmation — repo-lint 106/106 unique (files 198), queue [0.4.101] pointer-only 111 items/59 sections/64 bullets, suites 420+218+86 green against HEAD, committed md5 1f3f4213 == worktree, regression 54/4618/0, pushed b3ad629..2808a8c
> build · mimo-v2.6-flash-free
$ cd /data/repo && git status --porcelain; echo "--- BUDGET:"; grep -i "month\|spent\|remaining\|allowance" BUDGET.md | head -5; echo "--- unread:"; sqlite3 /opt/startup/dev/data/messages.db "select count(*) from messages where direction='investor_to_agent' and read=0;"; sqlite3 /opt/startup/prod/data/messages.db "select count(*) from messages where direction='investor_to_agent' and read=0;"; ./tools/inbox-status >/dev/null 2>&1; echo "inbox-status rc=$?"; echo "--- INBOX tail headings:"; grep -c "^## " INBOX.md; grep "^## " INBOX.md | tail -1
--- BUDGET:
month: 2026-09
allowance: 5.00
spent: 1.50 (gladex.de — registered + DNS live since 2026-09-20, 1.50/mo)
remaining: 3.50
  - 2026-09-24: ledger narrative corrected — spent-line still said "purchase pending at registrar" 4 days after registration (queued candidate #1); numbers unchanged (1.50/3.50, no money moved), log history above kept verbatim.
--- unread:
0
0
inbox-status rc=0
--- INBOX tail headings:
74
## 2026-09-27 investor: finish Immich setup via API NOW (no waiting)  ~~HANDLED 2026-09-27T18:22:44Z~~
> build · mimo-v2.6-flash-free
**Run complete — STEP 0 answered first, then one verifiable step, committed and pushed.**

## STEP 0 (before anything else)
Measured first, not assumed: `investor_to_agent` **unread = 0 dev / 0 prod** (direct `SELECT count(*)`), `inbox-status` → **exit 0, owed 0** (74 entries / 73 handled / 1 open — line 468, the Nextcloud half of the account ask, already **replied** and blocked on REPORT.md §14). No order was waiting, so I posted the run's status reply parameterised into both DBs: **dev 97 / prod 63**, `read=1`; re-verified unread still **0 / 0**.

## The step — queue item (46) `[0.4.102]`
`regression-run --note` shipped unbounded while `print_baseline` prints it behind a **fixed two-space prefix** — a newline in a note became an *unprefixed* second line that reads as the tool's own output.

- `NOTE_MAX = 200`, one constant; `note_is_blank()` / `note_problem()` are now the single blankness and bound rules for **both** halves.
- Write side: control character → exit 2 `must be a single line …`; over length → exit 2 `must be at most 200 characters (got N)` — before any suite runs, JSON refusal still one 14-key object, stderr empty.
- Read side: `load_baseline` holds a **stored** note to the same two bounds, at the `format`/`kind`/`totals` position.
- `--help` and the docstring both interpolate the constant (A10–A12 pin the single definition and both renderings).

**Measured:** `test_regression_run.sh` **218 → 237 / 0** (19 new). **Pre-fix replay** in place (old md5 `1f3f4213…`, restored and hash-verified) → **222 / 15**, every red a finding, 4 self-identifying guards. **Regression on the frozen tree, re-taken after the doc appends:** **54 suites / 4637 / 0 / 0, exit 0** = baseline 4618 **+ exactly 19**. Two of my own first-draft defects recorded rather than hidden (A11 vs argparse's help wrapping; an empty derived `note_max` producing reds about the empty string).

**Also found and fixed:** the previous queue bullet said "carry … unchanged" but re-listed only the older list, silently dropping items **(44)** and **(45)** — restored with their original wording.

**Gates:** repo-lint **107/107 unique** (files 198, `citations_missing []`), queue-source-check **exit 0** (111 items frozen / 60 sections / 65 bullets), source-sync-check **in sync 42/2**, suites 420+237+86 green against HEAD, committed md5 == worktree, `system-status` overall ok / errors 0 (same three standing warnings).

**Committed & pushed:** `768e5a6` (step) + `40f2a4d` (post-commit confirmation), staged **by explicit path** (never `git add -A`), `bf72efa..40f2a4d` on `origin/main`, tree clean. New queue item **(47)** filed: nothing compares the tool's exit-2 list with `REGISTRY.md`'s prose copy of it.

**Safety:** model spend **0.00** (`*-free` only), `BUDGET.md` untouched (1.50 / 3.50), **zero DNS writes**, no API key, no secret read, no service restarted, no mail sent, Nextcloud still `installed:false` (§14 open), Immich library `total: 0` / 8 accounts.
exit=0

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

[Mon Sep 28 12:09:47 2026] 127.0.0.1:51844 Accepted
[Mon Sep 28 12:09:47 2026] 127.0.0.1:51844 Closing
[Mon Sep 28 12:10:47 2026] 127.0.0.1:40130 Accepted
[Mon Sep 28 12:10:47 2026] 127.0.0.1:40130 Closing
[Mon Sep 28 12:11:47 2026] 127.0.0.1:56502 Accepted
[Mon Sep 28 12:11:47 2026] 127.0.0.1:56502 Closing
[Mon Sep 28 12:12:47 2026] 127.0.0.1:54546 Accepted
[Mon Sep 28 12:12:47 2026] 127.0.0.1:54546 Closing
[Mon Sep 28 12:13:47 2026] 127.0.0.1:48154 Accepted
[Mon Sep 28 12:13:47 2026] 127.0.0.1:48154 Closing
[Mon Sep 28 12:14:47 2026] 127.0.0.1:34470 Accepted
[Mon Sep 28 12:14:47 2026] 127.0.0.1:34470 Closing
[Mon Sep 28 12:28:41 2026] 127.0.0.1:44172 Accepted
[Mon Sep 28 12:28:41 2026] 127.0.0.1:44172 Closing
[Mon Sep 28 12:28:42 2026] 127.0.0.1:44182 Accepted
[Mon Sep 28 12:28:42 2026] 127.0.0.1:44182 Closing
[Mon Sep 28 12:30:54 2026] 127.0.0.1:59176 Accepted
[Mon Sep 28 12:30:54 2026] 127.0.0.1:59176 Closing
[Mon Sep 28 12:31:54 2026] 127.0.0.1:41156 Accepted
[Mon Sep 28 12:31:54 2026] 127.0.0.1:41156 Closing
[Mon Sep 28 12:32:54 2026] 127.0.0.1:60436 Accepted
[Mon Sep 28 12:32:54 2026] 127.0.0.1:60436 Closing
[Mon Sep 28 12:33:54 2026] 127.0.0.1:47408 Accepted
[Mon Sep 28 12:33:54 2026] 127.0.0.1:47408 Closing
[Mon Sep 28 12:34:54 2026] 127.0.0.1:49594 Accepted
[Mon Sep 28 12:34:54 2026] 127.0.0.1:49594 Closing
[Mon Sep 28 12:35:54 2026] 127.0.0.1:38868 Accepted
[Mon Sep 28 12:35:54 2026] 127.0.0.1:38868 Closing
[Mon Sep 28 12:36:54 2026] 127.0.0.1:60072 Accepted
[Mon Sep 28 12:36:54 2026] 127.0.0.1:60072 Closing
[Mon Sep 28 12:37:54 2026] 127.0.0.1:53554 Accepted
[Mon Sep 28 12:37:54 2026] 127.0.0.1:53554 Closing
[Mon Sep 28 12:38:54 2026] 127.0.0.1:45994 Accepted
[Mon Sep 28 12:38:54 2026] 127.0.0.1:45994 Closing
[Mon Sep 28 12:39:54 2026] 127.0.0.1:53700 Accepted
[Mon Sep 28 12:39:54 2026] 127.0.0.1:53700 Closing
[Mon Sep 28 12:40:54 2026] 127.0.0.1:54316 Accepted
[Mon Sep 28 12:40:54 2026] 127.0.0.1:54316 Closing
[Mon Sep 28 12:41:42 2026] 127.0.0.1:38460 Accepted
[Mon Sep 28 12:41:42 2026] 127.0.0.1:38460 Closing
[Mon Sep 28 12:41:42 2026] 127.0.0.1:38470 Accepted
[Mon Sep 28 12:41:42 2026] 127.0.0.1:38470 Closing
[Mon Sep 28 12:41:43 2026] 127.0.0.1:38478 Accepted
[Mon Sep 28 12:41:43 2026] 127.0.0.1:38478 Closing
[Mon Sep 28 12:41:46 2026] 127.0.0.1:59848 Accepted
[Mon Sep 28 12:41:46 2026] 127.0.0.1:59848 Closing
[Mon Sep 28 12:41:47 2026] 127.0.0.1:59854 Accepted
[Mon Sep 28 12:41:47 2026] 127.0.0.1:59854 Closing
[Mon Sep 28 12:41:47 2026] 127.0.0.1:59864 Accepted
[Mon Sep 28 12:41:47 2026] 127.0.0.1:59864 Closing
[Mon Sep 28 12:41:51 2026] 127.0.0.1:59866 Accepted
[Mon Sep 28 12:41:51 2026] 127.0.0.1:59866 Closing
[Mon Sep 28 12:41:52 2026] 127.0.0.1:59874 Accepted
[Mon Sep 28 12:41:52 2026] 127.0.0.1:59874 Closing
[Mon Sep 28 12:41:52 2026] 127.0.0.1:59890 Accepted
[Mon Sep 28 12:41:52 2026] 127.0.0.1:59890 Closing
[Mon Sep 28 12:41:54 2026] 127.0.0.1:59894 Accepted
[Mon Sep 28 12:41:54 2026] 127.0.0.1:59894 Closing
[Mon Sep 28 12:41:56 2026] 127.0.0.1:59634 Accepted