# REPORT — Securo onboarding, 2026-11-15 (attended) > A sibling of `REPORT.md` (another session owns that file today). ## In plain words - **Securo is in the live catalog** (catalog `b605815`, on the operator's yes). Every box offers it within 15 minutes; 9202 offers it now (read back from the live catalog). - It went through all 63 checks of the new-app checklist: bench and scratch box 9202, the update step 0.17.0 → 0.18.0, the undo on a real failure, the setup gate against a stranger, sign-in, lock-out, family members, a receipt upload, and backup → remove → restore. Every row is done or n/a (`app-catalog-felhom.eu/onboarding/securo.md`). - **Bank sync is optional and not set up by the box.** The household loads the file from its online bank. - **Website:** the card pictures were published by the operator's run; the Securo card is on the apps page in both languages and every app count reads 64 (R-1399 closed; wger is hidden on purpose and needs no card). The card goes public with the next publish run. ## Commits and CI by commit | Repo | Commit | What | CI job | |---|---|---|---| | felhom.eu | `44715fb3` | website assets: Securo's logo + three screenshots | 2695 success | | felhom.eu | `c616727d` | the audit evidence + R-1399 | 2697 success | | app-catalog-felhom.eu | `b6058150` | the template, its record, fixture, data-key reason, copy freeze, README, FIRST-ADMIN | 2698 success | | app-catalog-drill | `99ef45e` … `35e651a0` | the drill steps of the walks (FROM, bumps, the undo case and its reverts) | — | ## Register Rows at this session's first register read: 249; now 252. This session opened 2 (R-1399 the website app list, R-1402 the catalog's decoy suite) and closed none; session 1 opened R-1400 and R-1401 and closed R-1397 in the same time. ## Observations - The website lists fewer apps than the catalog — FILED: R-1399, then FIXED the same day (Securo's card, the counts at 64; wger is hidden on purpose). - `scripts/test_gate_decoys.py` fails 17 decoys of the move gate and the security-fix checks on a CLEAN clone of catalog main `f4c5c79` too (my changes add none) — FILED: R-1402. - Securo starts in English until each browser picks a language — NOT-A-FINDING: upstream's choice; the first steps say so. - Every LAN visitor shares one sign-in limit (5 a minute) — NOT-A-FINDING: the internet visitors each get their own. - The 0.17.0 image refuses the 0.18.0 schema (an abort cannot run old code on migrated data) — NOT-A-FINDING: the box's undo restores the data and was proven. - The first box walk was `inconclusive`: the fixture used `@example.invalid`, which Securo's e-mail check refuses — fixed in the fixture (small). - The first lock-out measurement was a tool error (a heredoc newline in the simulated tunnel's body) — fixed, measured again; only the second run is evidence. - The website publish dry run showed the installer pin moved today (another session) — checked: the live installer is the same file (sha256 `93a021db…`), so a publish changes only the four Securo pictures. ## Teardown Machine: 9202 back on the live catalog (controller.yaml restored byte-identical, my copy of it removed), no Securo container or volume left; the drill mirror stays in 9202's volume (harmless, as after B23). Bench 9401 stopped (as found), its Securo scratch files removed; the bench's `templates/securo` and fixture file now equal the catalog. Host: nothing. Hub: nothing. DooPlex: the 9202 password file and session files shredded/removed. Evidence: `documentation/audits/new-app-securo-2026-11-15/`.