51 KiB
AI Agent Note — VardKey
Short working handover for the next model. Keep it short.
TODO.md(Norwegian) is the requirement specification. Sections 1–18 are the binding spec; 19–31 are history; section 32 is the round in progress. The full reasoning is in the git history.- The binding rules live in
TODO.mdandREADME.md. This note does not repeat them — it carries the handover, the rules that are easy to get wrong, and the traps that already cost time. HANDOVER.mdandai-thought.mdwere deleted as outdated. Do not recreate either.
Status
- Done and verified: sections 1–31 and section 18 steps 1–11, plus section 32.1 (the icon), 32.2 (the UI pass and the retro theme), 32.3 (the alert and page texts), 32.4 (the bug report, rounds 32-A and 32-B) and 32.7 (the top-bar menu and the dashboard clean-up), each with its follow-up round, and then 32.8 stages A, B and C (the login lockout, the date limit, the sticky top bar, the package task, the new key row per rotation with the frozen history, and the security source taken from the row). Schema version 13. 532 green tests. The StartOS package is built, sideloaded and running on a StartOS 0.4.0 box; the package version is still
0.1.0:0. - What is left: 32.8 point 7, the user's manual end-to-end pass (BR-003), which no agent can do; then 32.6 the documentation drift, and then 32.5 the raise to
1.0.0:0together with section 18 step 12. 32.6 has to come after the rounds it documents, and 32.5 last. The user's manual pass over the finished §32 work is outstanding as well, and it is what point 7 is about: read the note under 32.4 before acting on it, because the report's own wording for it is out of date. - A round stops when it is done. Do not start the next item unless the user asks.
- Verified how: the automated suite, the user's manual browser testing, and the install and test round on a StartOS 0.4.0 box. The user runs the suite; the agent has no shell.
- Tests:
python3 -m unittest discover -s tests. They run on a Windows copy atC:\Users\<brukarnamn>\Downloads\helsesjekk, so they must pass there as well as on Linux. Stage C went from 525 to 532 tests, and the user ran the suite: it is green on both platforms. The519figure in the stage-B line ofTODO.md§32.8 was written in stage B and was not updated when C1's migration tests landed, which is where the six tests between 519 and 525 come from. - Keep this note short. The finished §32 sub-rounds are one short line each in
TODO.md§32, and the reasoning is in the git history. A finished round does not need its narrative back, and the two documents should not be rewritten in full every round.
Section 32 — what is left
- 32.8 The seven points before 32.6 and 32.5. Listed with checkboxes in
TODO.md§32.8, in three stages. Stage A is points 3, 4 and 2: no check date in the future, a top bar that sticks to the top, and a login lockout (five failures, fifteen minutes, the new429). Stage B is point 1: the package holds the service until the admin password is set, through a required StartOS task that runsmaintenance.py set-admin-password— the password is still generated inside the container, and the web create-admin form stays as the local bootstrap. Stage C is points 6 and 5 together, because both need schema 13:ai_security_checks.key_idso a security alert's source can never flip on a rename, software-sourced closures listed in the history and the CSV, and a rotation that inserts a newkeysrow with the samekey_numberplus a name/type/model snapshot on every check, so older history can never change. Stages A, B and C are done; only point 7 is left, which is the user's manual end-to-end pass (BR-003) and cannot be done by an agent. - 32.6 The documentation drift.
README.md,docs/application.md,instructions.md,UPDATING.md,AGENTS.mdand the package's own i18n strings are all behind the code, andTODO.md§32.6 lists what is wrong in each. It answers the rest of BR-004, and it must be done before 32.5, so the documents describe the released state. In one line, what the documents still miss: the login and session endpoints and the gating/CSRF rule, the CSV formula guard, the canonical path comparison, the local secret guard, the two summary fields, the moved field reminder, the addedmodelfield in the check detail, the icon link and the two PNGs, the top-bar menu, the removed dashboard cards, the export section and its note on the history page, the#history-detailanchor, and the robots rule. After 32.8 the schema is 13 and the documents also miss the admin step and the package task, the429lockout, the date limit, the sticky top bar, the new key row per rotation with a frozen history, the snapshot as the source of the name and the type on a check, and software closures in the history and the CSV. - 32.5 The version raise.
0.1.0:0→1.0.0:0, once 32.6 is done and the user says so. No schema change, no new endpoint, no new feature.startos/versions/stays empty, because the app migrates its own schema.
BR-003 — the one open item in 32.4
The report says the manual end-to-end pass is still open, but it quotes a §31 text that no longer exists, and §16 and §17 record that the manual browser pass and the box tests were run and verified through §31. What is outstanding is the new manual work §32 added — the icon, the retro theme on a narrow phone, the alert texts, the top-bar menu, the cleaned dashboard, the export section on the history page, the PWA with the two PNG icons, the six finished points in §32.8, and the whole box test again after 32.5 — plus the user's confirmation that it has been run. Do not re-verify what §16 and §17 already record as done. The other verdicts are settled: BR-001 is a false positive, BR-004 falls away, BR-005 is verified with no behaviour change, and BR-002, BR-006, BR-007 and BR-008 are fixed with regression tests. The report was written against a different file set and asks for eight documents that do not exist here; do not create them.
The wallet-secret guard (round 32-B)
secret_guard.py is the one place that decides whether a piece of text looks like a wallet secret. It is local (no network call), silent (the text never reaches a log, a message or an exception) and narrow (it matches only the shapes that are unmistakable, so an ordinary note is not refused).
- What it recognises: a PEM private-key block; an extended private key with any SLIP-132 prefix (
xprv,yprv,zprv,tprv,uprv,vprv) at 100 base58 characters or more; a WIF key (51 or 52 base58 characters starting5,9,K,Lorc); a BIP38 key (6P, 58 characters); and a run of 12 or more consecutive BIP39 words, also in the middle of a sentence. It returns a short English label such as"a seed phrase", never the value, andrejection_message(field, label)never repeats it either. - The wordlist is embedded, 2048 words, from
bitcoin/bipsbip-0039/english.txt.tests/test_secret_guard.pychecks it: 2048 words, alphabetical, 3–8 lower-case letters, and no two words sharing their first four letters. That last check is the strong one — the specification guarantees it, so a mistyped or missing word breaks it. Do not edit the list without that test. - Where it is wired in:
validation.pycalls_reject_secret()from the four text helpers_text,_key_name,_modeland_required_path, so every free-text field is covered by construction.auth.pychecks the admin username the same way; the password is deliberately not checked, because a long passphrase is a legitimate password and refusing one would lock the user out. - The refusal is a
ValidationError(SecretDetected), so the existing handlers answer400.app.pyneeded no change at all — do not add a handler for this. A plainValueErrorwould have been caught bySTORAGE_ERRORSand answered500. - The guard runs before the field's length limit. A seed phrase pasted into a 60-character label field is answered with "looks like a seed phrase", not with "too long" — the other order would invite the user to shorten a seed phrase instead of removing it. The limit is still the outer bound. This order is deliberate and pinned by tests; do not swap it back.
- The reminder in the UI is one line, in the key setup — a single
p.secret-warningabove the whole key list instatic/setup.html, and nowhere else; the per-field copies were removed in 32.2. The page notice is a different thing and stays onhealth-check.html,setup.htmlandsettings.html. The server side is unchanged. - The limitations are documented in the module docstring and are not to be played down: a bare 64-character hex string is not refused (it is indistinguishable from a transaction id, which is legitimate note content); only the English wordlist is used; a secret split across several fields, or in a shape outside the patterns, is not caught; and a false positive is possible for a note that happens to be twelve or more BIP39 words in a row.
Alerts, closures and key identity — the parts that are easy to get wrong
The rules are in TODO.md §7; these are the traps that have already cost time.
- The acknowledgement stores the key.
alert_acknowledgements.key_idis nullable and holds the key the alert belonged to, orNULLfor a software alert raised by the settings list. An alert with a key is acknowledged and hidden by(code, key_id); an alert without one stays on(code, source). This closed a real bug: a rename — or arotatewithkeep_name=false— used to reopen a closed alert, becausesourcecarries the name the key has now. - A key-sourced security alert carries
Key N: <name>, and the row decides.build_alerts()reads the key checks into a lookup by key id, and_security_alerts()looks the row's ownkey_idup in it. When the row carries the checked key, the key is the source — the sameKey N: <name>label every other key alert uses — and the alert belongs to it, so it is closed by a rotation; otherwise the source is the plain software name and it is closed by removing that software from the settings. The provider name is never compared with the names that are configured now, so a rename or a rotation cannot turn a key alert into a software one (section 32.8 point 6).rotation_alert_targets(),software_alert_targets()andnewest_alert_check_id()all read the samebuild_alerts(), so they see the same source, which is what keeps an alert closable. - The contact alert belongs to the checked key. It gets
key_idandsource=Key N: <name>, and its message says "this key" and does not repeat the name. The severity staysinfoand the action stayscontact_updated. It blocks its key like every other key alert, becauseblocking_alert()and theblockingAlertsfilter in the form only look atkey_id. A record without a key check raises no contact alert at all. software_incompatiblehas two forms. A software from the settings list keepssource= the software name,key_id=None, actions[remove_software], and carriessoftware. A wallet software that is the name of a configured cloud key can never be removed from the settings, so that form belongs to the checked key instead:key_id= the checked key,source=Key N: <name>, actions[replace_hardware], and nosoftwarevalue, so a settings removal can never close it. Only that form names the software in its message, because its source is the key.alerts.pystill needs the cloud key names, but only forsoftware_incompatible.build_alerts()andcollect_alerts()takecloud_names=(), threaded throughrotation_alert_targets(),software_alert_targets(), and the four call sites inapp.py(_send_alerts,_blocking_alert,_rotate_key,_close_removed_software), plusnewest_alert_check_id().app.cloud_key_names(keys)builds the tuple. Ifnewest_alert_check_id()does not get exactly the same names ascollect_alerts(), the alert'ssourcediffers, the acknowledgement storescheck_id = None,_is_acknowledged()refuses to hide an alert without a check id, and the alert can never be closed. Keep the default()onbuild_alerts()._send_health_check_list()passes the same tuple for itsneeds_follow_upfield, for the same reason. Section 32.8 point 6 removed the name-matching half of this trap for security alerts: eachai_security_checksrow carries its ownkey_id, so the source no longer depends on the current names.- Every closure of a check is shown, key-sourced or not.
app.closed_alerts()anddatabase._grouped_closures()list every acknowledgement that carries acheck_id, so a rotation, a replaced hardware, an updated contact, an ignored key alert, and a software alert closed by removing it are all shown; only an acknowledgement without a check id is left out.GET /api/health-checks/<id>returns them asclosed_alerts, and the CSV has one last column,Follow-up closed, withcode: action (YYYY-MM-DD), hanging on the check it closed throughcheck_id. The history table showsAlert,ActionandClosedonly — the source is not a column. The filter onkey_id IS NOT NULLwas removed in section 32.8 point 6, on purpose. - The key card shows the next due date with the alert note.
createKeyCard()instatic/app.jsalways rendersNext check dueand adds the alert note on top.Last checkedwas dropped in 32.2, but the note still comes in addition to the date, never instead of it — that is what cost the dates once before, for a key whose only alert was the informational contact one. - The type is part of a key's identity. The setup locks it, the settings never change it, and only a rotation may pick the other one:
POST /api/keys/<id>/rotatetakes an optionaltype, and leaving it out keeps the current type.save_key_settings(..., key_type=None)usesCOALESCE(?, type), so the settings and a rotation share one statement — but since section 32.8 point 5 a rotation does not use it:rotate_key()inserts a new row for the samekey_numberand retires the old one, so the settings write the active row in place and the rotation does not. - A key that becomes a cloud key must not take a taken name. A cloud key offers its own name as a wallet software choice, and the settings list holds the other choices, so a name the list already holds is refused with
409throughWalletSoftwareNameInUse— aValidationErrorsubclass, so it has to be caught before the generalValidationErrorin_rotate_key, or the409turns into a400.validate_rotation()therefore takes the wallet software list as a fourth argument. - A rename needs no rotation and no confirmation. In the rotation section the name can be changed on its own through
POST /api/settings/keys, for any key, and the address path of a key with no saved check can be changed there the same way. Marking a key withRotate this keyand ticking the confirmation box sends it through the rotate endpoint instead, together with the type and the new path. The single button isSave keys.
Deliberately left out
remove_software is still a dead end in two places: a software-sourced security_vulnerable offers it next to ignore (so the alert stays closable, but one button leads nowhere), and a software-sourced security_compromised offers it as its only action, so an alert whose "software" is a cloud key name can never be closed. Neither is fixed.
Rules that are easy to get wrong — do not re-open
- One submission records one key. There is no "round" grouping, and each key's next check is its own date plus one calendar year, computed on the server and not editable in the form.
- A key that could not be checked records nothing else. That covers
available=0and a recorded address path that does not match.wallet_softwareandai_security_checksmust be empty; a non-empty value is a400, not something dropped silently. Nothing is stored for them, andsoftware_compatibility_statusbecomesNOT_ASSESSED. Availabilityis required and offers only Yes/No. A cloud key collects onlyAvailability; its compatibility is derived from it, and its own name is the wallet software.- A hardware key needs an explicit compatibility status, and
Hardware worksneeds an explicitYesorNo. A missing value,""andNOT_ASSESSEDare all a400. - The contact confirmation is required for every submission, including one for a key that could not be checked, and it has no "not assessed" value.
NOT_ASSESSEDis not offered or accepted anywhere else. No field but the derivedsoftware_compatibility_statusaccepts it.Address path verifiedis never a form field again. The server derives it from the comparison.- The address path is locked after the first saved check, and no two keys may share one — in the setup, in the settings or in a rotation. The comparison is canonical:
m/48h/0h/0h/2handm/48'/0'/0'/2'are the same path, and the stored spelling is never rewritten. - The type is locked by the setup, and only a rotation may change it. The settings never take a
type, and a rename alone is not a rotation. - An alert is closed by the user, not by a newer check. A newer check that raises nothing does not clear an older alert, and a key with an open alert cannot get a new check at all.
- An acknowledgement stores the action name, not a free reason, and only the actions the endpoint performs are accepted.
rotatehas its own endpoint, because it requires a new address path. - Every alert message says what is wrong, which key or software, and what closes it, and it does not repeat the name in
source— except the cloud-key form ofsoftware_incompatible. SeeTODO.md§7. - The export is plain CSV only. XLSX is not built and the app does not promise Excel compatibility. The column names and the column order are an agreement; do not change them. A text cell that a spreadsheet would read as a formula is neutralized with a leading apostrophe, and that is the only change a cell may receive.
- The PWA cache stores only a valid shell answer.
isCacheable()instatic/sw.jsrequiresresponse.ok,response.type === "basic"and!response.redirected, so a login page (or an error page) can never replace the cached shell. The allow-list is the other half: the worker never names/api/*,/health,/loginor/robots.txt. - No page is meant to be crawled. All six pages carry
<meta name="robots" content="noindex, nofollow">in<head>,_send_security_headers()inapp.pyaddsX-Robots-Tag: noindex, nofollow, andstatic/robots.txtanswersDisallow: /. A new page has to carry the meta, andstatic/robots.txtmust not be added to the service worker's shell list. The header is not sent bysend_error(), so a404and a500answer without it — deliberate, because the round asked for the pages, the JSON, a redirect, the CSV and the static files. - The navigation lives in the top bar, not on the page.
static/session.jsbuilds the menu, the panel and the.topbar-actionswrapper on the five application pages;/loginhas none. The dashboard markup therefore carries nohreffor/health-check,/settings,/historyor/api/export, and the export link and its note are on the history page. The page the user is already on is a<span>witharia-current="page", not a link. Do not add a navigation link back to the dashboard markup:tests/test_app.py'stest_the_dashboard_carries_no_navigation_link_of_its_ownpins it. - The Attention card is the only summary card. Its detail line carries the overdue list when something is overdue and the earliest due date otherwise;
nextCheckTitleandnextCheckDetailno longer exist, and.overview-gridis one column at every width. - An empty alert list renders nothing. The count beside the heading says
No alerts; do not put placeholder text back into#alerts-list. - The two summary fields are additive and mean what they say.
key_namesfollowskey_numbersand carries the name the check was recorded with, read from the snapshot on the check, so a rotation or a rename cannot rewrite an older summary;needs_follow_upis a property of the saved check, so it does not change when an alert is closed or a key is renamed. Do not "fix" it to followcollect_alerts(), which hides acknowledged alerts and would rewrite history. - The prohibition lists in sections 1 and 13 of
TODO.mdstay unticked. They are standing prohibitions, to be re-confirmed before release — with the one amendment for the installable PWA. - Whether a date is overdue is decided in the browser, so
/api/key-statusstays independent of the current date. /api/previous-valuesstill returns asecuritymap that nothing consumes any more, and itsaddress_pathper key is no longer used by the form, which pre-fills the path fromkeys.address_path. Both are deliberate: removing them would shrink the endpoint, but nothing breaks while they stay.delete_all_data()deletes the checks before the keys.key_checks.key_idisON DELETE RESTRICT, so the other order fails. The acknowledgements are deleted explicitly, because they have no foreign key at all. The order is unchanged by section 32.8 point 5, and the delete has to clear every row ofkeys, active and retired alike.
Data model (schema version 13)
keys:key_number(1–multisig_key_count),name,type(hardware/cloud),model,address_path,active,rotated_at,created_at. There is nodevicecolumn. The name is editable in the settings; the type and the address path are the identity of the key, the path must be unique across the active rows, and the type only changes through a rotation.modelis optional documentation and is not locked.activeis1on the row the app reads as "the key", androtated_atrecords when a row was retired.health_checks:checked_at,next_check_due,wallet_software,software_version,software_compatibility_status,software_comments,contact_info_verified,comments, plus the legacysparrow_version/nunchuk_version.wallet_softwareandcontact_info_verifiedare free text.key_checks: one row per submission —key_id,key_name,key_type,key_model,available,hardware_works,firmware_version,seed_backup_verified,address_path_verified,address_path,comments. The two cloud columns are gone. The three snapshot columns are a frozen copy of the key the check was recorded for.ai_security_checks:provider(free text),key_id(nullable),status,comments, withUNIQUE (health_check_id, provider).key_idis the checked key when the provider is that key's own configured name, andNULLotherwise.wallet_software:name,created_at. Stored whole, seeded withDEFAULT_WALLET_SOFTWARE(Sparrow Wallet) when the list is empty, never emptied through the API.delete_all_data()empties it, and the seed puts the default back.alert_acknowledgements:code,source,key_id,check_id,reason,acknowledged_at.reasonholds the action name that closed the alert, andkey_idholds the key the alert belonged to, orNULLfor a software alert from the settings list.app_settings:name(PRIMARY KEY),value. Holdsmultisig_threshold,multisig_key_count,admin_username,admin_password_hash,session_secret,failed_login_countandfailed_login_until.delete_all_data()removes the two multisig rows, so the 2-of-3 default applies again, and keeps the credential rows and the two lockout rows.tool_checksno longer exists.- Startup migration rebuilds
keyswhen it still hasdevice, the oldkey_numberCHECK, or noactiveflag;key_checkswhen it uses the old layout, still holds the two cloud columns, or still allows the old values; andhealth_checkswhen itsCHECKconstraints still allow the old values. It then dropstool_checksif it is there, adds the newer columns throughOPTIONAL_COLUMNS, backfills the snapshot on every check from the key it belongs to, and creates the partial unique index over the active row. Schema 13 is done (section 32.8 points 5 and 6): theUNIQUEonkeys.key_numberis a partial unique index over the active row (idx_keys_active_key_number … WHERE active = 1),keyscarriesactiveandrotated_at,key_checkscarries the name/type/model snapshot, andai_security_checkscarries a nullablekey_id. A row written before the flag is active, and a security row written before the column keeps aNULLkey. - The table definitions are
{table}templates, so a rebuild can create the same table under a temporary name.health_checksis a parent table, so the whole migration runs withPRAGMA foreign_keys = OFF, checksPRAGMA foreign_key_checkbefore committing, and rolls back on any failure. tests/test_database.pycarriesLEGACY_SCHEMA,SCHEMA_VERSION_5,SCHEMA_VERSION_8,SCHEMA_VERSION_9,SCHEMA_VERSION_10,SCHEMA_VERSION_11andSCHEMA_VERSION_12as literal older schemas. Copy that pattern for a new migration test: write the old schema, runinitialize_database(), then assert on the migrated rows. The schema-13 migration has its own test there, with the literalSCHEMA_VERSION_12, covering the snapshot backfill and the partial index.- There is no data migration for an older test database: the user deletes the
datafolder and configures the setup again. An acknowledgement written before version 11 has no key, and the older contact alert used the sourceContact information, which no longer matches, so the alert it closed is raised again. Section 29 renamed the database file for the same reason, and the old file is discarded with the rest.
API
Pages: /, /setup, /settings, /health-check, /history, /login. Endpoints: GET /health, GET /api/session, POST /api/admin, POST /api/login, POST /api/logout, GET /api/setup, POST /api/setup, GET /api/keys, GET /api/key-status, GET /api/previous-values, GET /api/settings, POST /api/settings/keys, POST /api/settings/wallet-software, POST /api/keys/<id>/rotate, GET /api/health-checks, GET /api/health-checks/<id>, GET /api/alerts, POST /api/alerts/acknowledge, GET /api/export, POST /api/health-checks. Field details are in docs/application.md — which is behind the code, see the gap under Status.
GET /api/health-checks returns each summary with id, checked_at, next_check_due, software_compatibility_status, wallet_software, contact_info_verified, comments, key_numbers, and the two added in 32.2: key_names (the name the check was recorded with, read from the snapshot) and needs_follow_up.
Maintenance commands, run inside the container by the StartOS actions and usable by hand: maintenance.py delete-all-data and maintenance.py set-admin-password. Neither is an HTTP endpoint.
The rotate payload is replaced + keep_name + name + type + address_path, and type is optional. A rotation inserts a new row and retires the old one, and the acknowledgements are written against the old key id, which is the row the alerts and their checks belong to. GET /api/health-checks/<id> also returns closed_alerts (every closure of the check), and each of its key_checks rows carries model as well — the field was added in the follow-up round of 32.1 so the history can show the model and the firmware version on one line, and the fields that were already there keep their names. The settings page has two sections, and POST /api/settings/keys is what a rename alone goes through. The #history-detail anchor and the ?id= query on /history are client-side only: they never reach the server, so no route changed for them.
Five things that are easy to break:
- New static files must be added to
STATIC_FILESinapp.py, or they return404. That includes/settings,/settings.js,/login,/login.js,/session.js,/sw.js,/manifest.webmanifest,/icon.svg, the two raster icons/icon-192.pngand/icon-512.png, and/robots.txt, which istext/plain; charset=utf-8and is deliberately absent from the service worker's shell list. /api/setup,/api/settings,/api/alerts,/api/key-status,/api/previous-valuesand/api/sessionare extra endpoints that are not in the minimum list in section 11.STORAGE_ERRORSinapp.pyincludesRuntimeError, becauselist_keys()raises it for a partial setup. Keep it there, or a broken database answers with a traceback instead of a500. It also includesValueError, sovalidate_admin_credentials()has to be called in its owntryand mapped to a400before the storage handler — and it is why a refused secret has to be aValidationErrorrather than a bareValueError.- The settings, rotation and acknowledgement endpoints keep the same error handling as the existing ones: a bad payload is a
400, an unknown key or alert is a404, a locked or duplicate address path is a409, and a broken database is a500.DuplicateAddressPathandWalletSoftwareNameInUsemust be caught before the generalValidationError— in_save_key_settingsand_rotate_keyrespectively — or the409turns into a400. A missing or invalid session is a401, a missing CSRF token is a403, and a locked login is the429from §32.8 point 2. newest_alert_check_id()inapp.pyderives the check id for the acknowledgement, the rotation and the software removal, so the same comparison rule is used in all three places.blocking_alert()is the module-level helper the409is built on, and_blocking_alert()reads the key id straight from the payload, so the rule applies before validation.
The StartOS package — the parts that bite
The package is one service: one daemon, one container, one SQLite database on the persistent main volume, mounted at /data. No Postgres, no Redis, no external backend, no extra container. The decisions are in TODO.md section 30.
- The application and the package share one repository, at the root. A deliberate deviation from the guide's "a package repo is not a fork of the application": the rule assumes a published upstream image or a git submodule, and VardKey has neither.
AGENTS.mdstates the deviation so nobody adds anupstream-project/submodule to "fix" it. - Anything the
Dockerfiledoes notCOPYis not in the image. The modules go in withCOPY *.py ./and the assets withCOPY static ./static;.dockerignorekeepstests,data,node_modules,javascriptand the build output out of the context.static/must be copied:app.pyanswers/healthfrom code, but every page is read fromROOT/static, so an image without it reports a healthy service whose whole UI answers404.static/robots.txtis served the same way, so it is in the image too. - Three things must agree.
startos/utils.tsholdsuiPort(8080);startos/main.tspasses it asPORTand uses it in the readiness URL;startos/interfaces.tsbinds it. Change one alone and the readiness check never goes green.DATABASE_PATHin the same file is what puts the database on the volume. The readiness check ischeckWebUrlagainsthttp://127.0.0.1:8080/health, and/healthmust stay open, or StartOS reports the service as unhealthy. A redECONNREFUSED 127.0.0.1:8080blip beforeVardKey listening on port 8080is just the first poll landing before Python has bound the socket; nothing to fix. - Two actions, registered through
Actions.of()instartos/actions/index.ts, with their strings instartos/i18n/dictionaries/:deleteAllData.ts(delete-all-data) andsetAdminPassword.ts(set-admin-password). Both runmaintenance.pyin awithTempsubcontainer, behind a confirmation the handler enforces. The generated password never travels into the container:maintenance.pygenerates it there and prints it, which is the packaging guide's own password-reset pattern. Section 32.8 point 1 adds a required task that holds the service until the admin password exists, instartos/init/requireAdmin.ts, and it keeps that same property: no password travels from the host into the container. - Backups.
startos/backups.tsissdk.Backups.ofVolumes('main'). StartOS stops the service for the duration of a backup, so the database file is quiescent while it is copied — no snapshot hook is needed, and it is not a known limitation. - No CI. The guide assumes GitHub Actions; this repository is on Gitea, so there are no
.github/workflows/and there must not be any. The.s9pkis built locally withmake. - The skeleton was written to match the Start9 template rather than generated by
s9pk init-workspace, because that command was not available where it was written. The successful build and install confirm they are equivalent;@start9labs/start-sdkis pinned to3.0.4. The task API used by §32.8 point 1 was verified against that pinned version (sdk.action.createOwnTask/createTask/sdk.action.clearTask,TaskSeverity, andcreateTaskansweringnull), becausenode_modules/is not in the repository. - The root
icon.svgis the package artwork, read from the repository root by the build. It is the same retro drawing asstatic/icon.svg, and it must stay under 40 KiB. - Documentation layout.
README.mdis the StartOS package README, with the guide's fixed heading set. The application's own reference isdocs/application.md, whichinstructions.mdlinks to.AGENTS.mdis the repo-specific agent context, andCLAUDE.mdis the one-line@AGENTS.mdimport. BothREADME.mdanddocs/application.mdare behind the code — that is 32.6.
Traps that already cost time — do not repeat
with sqlite3.connect(...)does not close the connection. It only commits or rolls back the transaction. On Windows the database file stays locked andTemporaryDirectory.cleanup()fails withWinError 32. Always close explicitly withtry/finally; seewrite_schema()intests/test_database.py.- A module the image is missing is a runtime error, not a build error.
auth.pywas once left out of theDockerfile, and bothapp.pyandmaintenance.pythen died withModuleNotFoundErrorinside the container while every test stayed green. TheCOPY *.py ./line is what stops that recurring. After any change to the image, check the modules, not just the tests.secret_guard.pyis covered by the same line. - Every server test module needs a session, and a redirect makes a missing one look like a pass. A page fetched without a session answers
302 → /login, andurllibfollows it, so an assertion on page markup reads the login page's HTML instead of failing with401. Usetests/server_session.py'sSessionMixin, and note that the manifest, the icons and/robots.txtare open while the pages are not. urlopenraisesHTTPErroron a 4xx. A test that expects a400has to catch it and read the body from the exception;tests/test_secret_guard.py'spost_status()helper is the pattern. A302behaves the same way when the opener refuses to follow it, which is howtests/test_app.pyreads a redirect's headers. A429is read the same way.- A field has a length limit, and the guard runs before it. A test that pastes an 80-character seed phrase into a 60-character label field gets a secret refusal, not a length refusal. Use a WIF key (52 characters) or a BIP38 key (58) when a test needs a secret that fits a short field, and use the seed phrase where the field has room.
- A
requiredcontrol inside adisplay:nonecontainer still blocks submission. The browser refuses to submit and logs "An invalid form field is not focusable". Hiding a section is not enough: the required flags must be cleared, or the section must render something other than required fields. That is whyrenderSecurityChecks()shows an empty note instead of cards when a key could not be checked, whysyncUncheckableState()clears the flags on the wallet software and the compatibility select — also for a cloud key, whose grid is hidden — and whysyncRequiredFields()clears the flag onhardware_worksandseed_backup_verified. The rotation fields follow the same rule: the type isdisableduntil the key is marked for rotation, the path isdisabledwhen the key has a saved check and is not marked, and the script refuses to send an empty path. The login page sidesteps it with two separate forms, so only the visible one is ever validated. - A hidden element has no box, so an anchor cannot jump to it.
#history-detailis hidden until a check is opened, so the browser's own:targetscroll does nothing.revealDetail()inhistory.jsunhides the section first and only then callsscrollIntoView(). Do not reorder those two lines, and do not drop the anchor from theViewlink or from the target insaveHealthCheck(). - A bare substring assertion over page markup also catches comments.
test_the_dashboard_has_no_next_check_card_and_no_history_blockfailed on its first run because it assertedassertNotIn("Next check", html)and the page carried an HTML comment naming the removed card. It now pins the markup that was actually removed —<p class="eyebrow">Next check</p>and the two ids. Assert on the element, not on the phrase; a phrase is also case-sensitive and would trip on any line such asNext check due. - A
display: gridpanel ignores thehiddenattribute..menu-panelis a grid, so the browser's own[hidden] { display: none }loses to it. The sheet carries.menu-panel[hidden] { display: none; }for that reason, the same way.form-stack[hidden],.field-grid[hidden]and.form-section[hidden]already did. Do not drop that rule. - The menu's outside-click listener has to allow the button's own click. The button's handler runs first and toggles the panel; the document listener then sees a target inside
.menuand does nothing. Reordering them, or replacingmenu.contains(event.target)with a check on the panel alone, makes the button close the panel it just opened. - The current-page check compares a normalised path, and the current item is a span.
currentPath()strips trailing slashes and keeps the root as/, so/health-check/matches/health-check. The match is rendered as a<span class="menu-item menu-item-current">witharia-current="page", never an<a>— that is what makes it unclickable. A test asserts the span, the class string and thearia-currentattribute, so switching back to a disabled link breaks it. - A sticky top bar needs a scroll container that is the viewport.
.topbaris a flex child ofbody, andbodysets nooverflow, soposition: sticky; top: 0works — but it stops working the moment an ancestor getsoverflow: hiddenorauto. The menu panel isposition: absoluteinside the bar, so it travels with it; keep itsz-indexabove the page content and below nothing that matters. - The mismatch confirmation follows its warning.
#address-path-confirmedisrequiredonly while#address-path-warningis visible, so the browser itself stops the submission and a required control is never hidden. - The per-field secret reminder is gone; only the setup keeps one. Since 32.2 it is a single
p.secret-warninginsetup.html, andsetup.js,health-check.jsandsettings.jsno longer build one at all. A test assertssetup.htmlholds exactly oneclass="secret-warning"and that the check form and the settings hold none, so re-adding a per-field copy breaks it. static/health-check.jsmust not useplaceholderanywhere. A test asserts the string is absent, because a previous value is a real input value.- The key card's line order is pinned by two test modules.
tests/test_key_status.pyassertscard.append(top, name)→card.append(model)→card.append(path), andtests/test_health_check_form.pyasserts that the top line is${String(entry.key_number).padStart(2, "0")} · ${keyTypeLabel(entry.type)}, thatcard.append(type)is gone, and that the model line is skipped bysameLabel(entry.model, entry.name). Adding a line, reordering them, or droppingsameLabel()breaks them. - The dashboard heading is dynamic and the static description is gone.
tests/test_app.pyassertsid="setup-shape"and the fallback text on the dashboard andassertNotIn("Bitcoin Multisig Health Monitor")there, whiletests/test_pwa.pystill asserts that exact description in the manifest. Do not move the one into the other. - The stylesheet test asserts on the shape of the theme, not on its wording.
test_stylesheet_is_the_retro_themechecks the two bevel colour pairs, the dotted focus outline, the font stack, the absence ofbox-shadowandgradient, and that everyborder-radiusin the sheet is0. It reads the values, so a restyled rule that keeps the palette passes, and a rounded corner does not. The onlyborder-radiusin the sheet is the explicit reset to0; do not drop it, and do not add a rounded corner anywhere. - The brand mark is an image, and the class name is pinned.
IconTestsintests/test_pwa.pyasserts that all six pages carry<img class="brand-mark" src="/icon.svg" alt="" aria-hidden="true">, that the old<span class="brand-mark" aria-hidden="true">V</span>is gone, that the sheet holdsimage-rendering: pixelated, and thatplace-items: centerappears nowhere. It also pinsmin-height: 100vh,min-height: 100dvh,flex-direction: columnand the exact stringmain { flex: 1 1 auto; }, which is what keeps the footer at the bottom. Do not style the brand mark as a letter again, and do not drop themainrule. - The two SVG files must keep the same drawing. The package icon at the repository root is the source of truth, and
static/icon.svgis the copy the PWA serves; only the leading comment differs, andtest_the_served_icon_carries_the_same_drawing_as_the_package_iconcompares them from<svg>onwards. A change to one drawing has to be made in both. - The two PNGs are real files, rasterized by the user, because the agent has no rasterizer.
IconTestsreads each file's IHDR and asserts 192×192 and 512×512, so a PNG made at the wrong size fails the suite instead of silently breaking the install prompt. - The manifest lists the SVG first, and the two PNGs after it.
test_the_manifest_lists_the_served_iconreadsicons[0], so a PNG added in front of the SVG turns the suite red. The PNG entries must carry the real size (192x192and512x512) andtype: "image/png". - The favicon link sits next to the manifest link, and
/loginhas no manifest.IconTestsasserts the exact string<link rel="icon" type="image/svg+xml" href="/icon.svg">on all six pages, so a reworded or reordered link breaks it./icon.svg, the two PNGs and/robots.txtare open (they are not inPROTECTED_PAGESand do not start with/api/), which is what lets the login page, the browser tab and a crawler read them. make x86runsprettier --check startos, and the config is the SDK's own..prettierrcis the single line"@start9labs/start-sdk/prettier.config.json", which is{ trailingComma: "all", tabWidth: 2, semi: false, singleQuote: true }— everything else is prettier's default, soprintWidthis 80 andquotePropsis"as-needed". That last one silently unquotes a single-word i18n key (Username: 21, not'Username': 21), and 80 columns decides when an array element or an argument collapses onto one line. Write it the way prettier would, or runnpx prettier --config node_modules/@start9labs/start-sdk/prettier.config.json --write startos.- The page heading uses
class="eyebrow"too. A regex overhealth-check.htmlfindsAnnual security checkbefore the numbered sections.settings.htmlstarts with theh1Settings, then01 / Wallet softwareand02 / Key rotation.tests/test_health_check_form.pyasserts those two lists exactly, so a new eyebrow on either page breaks it, and so does a removed one: update the list in the same change. The 32-B warnings deliberately usedclass="secret-warning", not an eyebrow. - A static help line that the script overwrites is dead text.
#software-helpinhealth-check.htmlwas emptied in round 32.3, becauselockSoftwareFields()fills it before the user can read it.static/health-check.jsmust keep setting it — the wording depends on the selected key. A test pins both halves: the empty element in the markup andsoftwareHelp.textContentin the script. aria-labelledbymay point at the eyebrow.login.html's first form has noh2any more; itsaria-labelledby="login-form-title"points at the eyebrow<p>that carries the sameid. Do not add a heading back just to have one, and do not drop theid.- Do not assert on the quoting style of the JavaScript.
static/health-check.jsuses single quotes for constant selectors and template literals where a${...}substitution is needed, andstatic/settings.jsuses double quotes. Assert on the selector text, e.g.[name="address_path"], and on quote-free substrings such as/api/settingsand/settings?rotate=. clean_record()intests/test_alerts.pyhas a fixed provider list and carries the key on the security rows. Eachai_security_checksentry is a dict withprovider,key_idandstatus: the entry whose provider is the key's own name carries the key number, and the others carryNone, which is what the storage writes. A test that needs another provider must replace the entry itself, and it has to keep thekey_idit wants, or the alert's source changes. The recorded software isSparrow Wallet, which is what a software alert names. A test that exercises the cloud-key form ofsoftware_incompatiblehas to passcloud_namestobuild_alerts(). The alert texts are asserted there, and not intests/test_app.py, which checks the code, the severity, the source, the actions and thekey_idonly — so a message change touchestests/test_alerts.pyalone.sample_payload()/unavailable_payload()are duplicated intests/test_validation.py,tests/test_database.py,tests/test_app.py,tests/test_maintenance.pyandtests/test_secret_guard.py. They build the labels from the key's name and the selected software, they always sendcontact_info_verified, and for a hardware keyhardware_works. They must send the key's ownaddress_path, because the server compares it with the stored one.- Every test fixture uses one address path per key (
ADDRESS_PATH,SECOND_ADDRESS_PATH,THIRD_ADDRESS_PATH), because the setup refuses duplicates.tests/test_app.pyalso hasROTATION_PATHSand therotate()helper, which picks an unused path per key so a test can rotate several keys in one run. With schema 13 a rotation retires the key row, so a test that rotates and then records a check has to use the new key id —rotate()hands it back underkey_id, read from the settings in its own answer. validate_health_check()takes a third argument (the wallet software list), andvalidate_rotation()takes a fourth for the same reason.tests/test_validation.pyandtests/test_database.pywrap the first one in a smallvalidate()helper.tests/test_app.pyhas seven test classes.ServerTestCasestarts the server and keeps the session and CSRF cookies a sign-in returns;AppHandlerTests,RenamedSetupTests,SettingsApiTests,RotationApiTestsandAcknowledgementApiTestsstore the default or renamed setup and callauthenticate();AuthApiTestsstarts from a database with no admin and covers the gating, the first-run admin, the login and logout, the CSRF check, the robots header on a redirect and the login lockout; andKeySetupApiTestscreates its admin throughPOST /api/adminand otherwise starts from an empty database. A new test must go in the class whose setup it needs. Every request helper sends the cookies and the CSRF header, andassert_status()is the one place an error status is expected.- The other server test modules share
tests/server_session.py.test_export_filename.py,test_key_model.py,test_key_status.py,test_multisig_setup.py,test_pwa.pyandtest_secret_guard.pymix inSessionMixin, which signs in through the API and carries the cookies; each class callsauthenticate()at the end of its ownsetUp. A test that switchesDATABASE_PATHto another file has to callauthenticate()again, because the session secret lives in the database, so the old cookie is signed with the wrong secret and every request answers401. The mixin resolves the path withapp.get_database_path(), so the admin lands in the file the server actually reads. - A login lockout test has to clear the counter.
AuthApiTestsshares one database per test, and the lockout lives inapp_settings, so a test that trips it leaves the account locked for the tests that follow unless it resets the two rows or uses its own database file. - The export link and its note live on the history page, and four tests moved with them.
tests/test_export_filename.pyreadsstatic/history.htmland asserts the dashboard has no/api/exportlink;tests/test_app.py's dashboard test becametest_the_dashboard_carries_no_navigation_link_of_its_own;test_the_history_page_offers_the_export_in_its_own_sectionintests/test_health_check_form.pypins theExportsection, the title bar, the note under the button and the absent.page-heading > .secondary-buttonrule; andtest_the_history_links_to_the_detail_sectionpins both anchors. A future round that adds a link to the dashboard has to update these, not delete them. #key-countstill exists insetup.html, but not on the dashboard. The dashboard's counter span was removed in 32.7;setup.htmlstill has theKey slotsinput withid="key-count"andname="key_count". The two are on different pages and were never the same element, so assert per file.- Section 19 records one key per submission, so a
Check idin the export can cover up to three keys only for rows saved before section 19. A legacy record with severalkey_checksis read through its first one when an alert needs the key it belongs to. - The database path default is repeated in sixteen places.
database.pycarriesdatabase_path="data/vardkey.sqlite3"as the default of every storage function (fifteen of them), andapp.pycarries it inget_database_path(). A missed one makes a single code path read or create the other file, silently. Grep for the old name before and after a rename round.
Product clarification
The key's configured name is its product label, and a cloud key's name is also a wallet-software choice. A hardware key name is never a wallet-software choice, and a cloud key records its own name as its software. A key is identified by its type and its address path, while the name is only a label that the user may change. The address path is unique per key, and a rotation replaces the key with a new one that must carry a new address path and may carry the other type. A rotation creates a new wallet, so the funds have to be moved from the old multisig setup to the new one before the old keys are discarded. An alert is closed by the user with the action that belongs to it, and a key with an open alert cannot be checked until the alert is closed. Since §32.8 point 5, a rotation also retires the old key row: the multisig slot number stays, the old checks keep the name and type they were recorded with, and the new row starts with an empty history.