Files
vardkey/ai-note.md
T

24 KiB
Raw Blame History

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; sections 19–29 are short summaries of the rounds that are implemented and verified, and section 30 is the StartOS package round. The full reasoning is in the git history.
  • The binding rules live in TODO.md and README.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  • HANDOVER.md was deleted as outdated. Do not recreate it.

Status

  • Done and verified: sections 1–30, and section 18 steps 1–11. Schema version 11. 346 green tests.
  • Section 30 is done and verified. The StartOS package lives in this repository beside the application: startos/, Dockerfile, .dockerignore, Makefile, package.json, tsconfig.json, .prettierrc, LICENSE, icon.svg, instructions.md, UPDATING.md, AGENTS.md, README.md, and CLAUDE.md. It built clean with make x86 into vardkey_x86_64.s9pk, was sideloaded onto a StartOS 0.4.0 box, and runs there: the service shows Running, the health check reports The web interface is ready, and Open UI opens the VardKey web app. Sections 14, 16 and 17 were exercised on that box and are ticked in TODO.md.
  • Two things are deliberately left in section 30, and both need the user's go-ahead. Do not raise the package version from 0.1.0:0 to 1.0.0:0, and do not draw the real icon.svg. Sections 16 and 17 are met, so section 30's own rule would bump the version — but the user has more to implement before version 1 and is taking that decision to their own AI strategist. Wait to be asked.
  • A round stops when it is done. Section 30 is the round and it is finished apart from those two deferred items. Do not roll into anything after it, and do not install or publish anything further.
  • Section 29 is done. The app is VardKey, described as Bitcoin Multisig Health Monitor, the database file is data/vardkey.sqlite3, and the export is vardkey-export.csv. The Gitea repo and the project id stay ai-bot/helsesjekk.
  • 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 manual tests in section 16 are all run.
  • Tests: the user runs them: python3 -m unittest discover -s tests. They run on a Windows copy at C:\Users\<brukarnamn>\Downloads\helsesjekk, so they must pass there as well as on Linux.
  • Future work, deliberately not built: more than three keys, and another threshold than 2-of-3.

Section 30 — the StartOS package

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 recorded in TODO.md section 30; this is what the next model needs on top of them.

Where the files live, and why

The application and the package share one repository, at the root. That is 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.md states the deviation so nobody adds an upstream-project/ submodule to "fix" it.

The image is built by the root Dockerfile, referenced as dockerBuild: {} in startos/manifest/index.ts, which makes the build context the repository root. Anything the Dockerfile does not COPY is not in the image. In particular static/ must be copied: app.py answers /health from code, but every page is read from ROOT/static, so an image without it reports a healthy service whose entire UI answers 404. The build log confirms four steps — FROM, WORKDIR, and the two COPY lines — and no USER step.

There is 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 .s9pk is built locally with make.

The three things that must agree

startos/utils.ts holds uiPort (8080). startos/main.ts passes it as PORT to the daemon and uses it in the readiness URL; startos/interfaces.ts binds it. Change one alone and the readiness check never goes green. The DATABASE_PATH constant in the same file is what puts the database on the volume instead of the application's own default of data/vardkey.sqlite3 relative to the working directory.

The readiness check is checkWebUrl against http://127.0.0.1:8080/health, not checkPortListening — section 30 asks for the endpoint, and /health is the application's own.

Backups

startos/backups.ts is sdk.Backups.ofVolumes('main'). The open question section 30 raised — whether setupBackups has a hook for a consistent SQLite snapshot — turned out not to need one: StartOS stops the service for the duration of a backup, so the database file is quiescent while it is copied. That is written up in README.md and in startos/backups.ts; it is not a known limitation. Restore was tested on the box and brought the setup and the saved checks back.

Notes on the skeleton

The skeleton files (package.json, tsconfig.json, Makefile, .prettierrc, startos/sdk.ts, startos/index.ts, startos/actions/index.ts, startos/i18n/, startos/fileModels/.gitkeep, assets/.gitkeep) were written to match the Start9 template rather than generated by s9pk init-workspace, because that command was not available in the environment that wrote them. The successful build and install confirm they are equivalent; @start9labs/start-sdk is pinned to 3.0.4 and that is the version the .s9pk was built with. No reconciliation is outstanding.

The icon.svg in the repository is a minimal placeholder, kept so the build has an icon to pack. Section 30 defers the real one to the 1.0.0 round.

Documentation layout

README.md is the StartOS package README, with the guide's fixed heading set. The application's own reference — the HTTP API, the alerts table, the schema history — moved to docs/application.md, which is what instructions.md links to under ## Documentation. AGENTS.md is the repo-specific agent context, and CLAUDE.md is the one-line @AGENTS.md import.

Sections 27 and 28 — implemented and verified

What the code does now, and the parts that are easy to get wrong:

  • The acknowledgement stores the key. alert_acknowledgements.key_id is nullable and holds the key the alert belonged to, or NULL for 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 a rotate with keep_name=false — used to reopen a closed alert, because source carries the name the key has now.
  • The contact alert belongs to the checked key. It gets key_id and source = Key N: <name>, and its message names the key: The contact details for Key 1: SeedSigner need an external update. Update them in your external documentation, then close this alert. The severity stays info and the action stays contact_updated. It blocks its key like every other key alert, because blocking_alert() and the blockingAlerts filter in the form only look at key_id. A record without a key check raises no contact alert at all.
  • software_incompatible has two forms. A software from the settings list keeps source = the software name, key_id = None, actions [remove_software], and carries software. 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 no software value, so a settings removal can never close it.
  • alerts.py needs the cloud key names. build_alerts() and collect_alerts() take cloud_names=(), threaded through rotation_alert_targets(), software_alert_targets(), and the four call sites in app.py (_send_alerts, _blocking_alert, _rotate_key, _close_removed_software), plus newest_alert_check_id(). app.cloud_key_names(keys) builds the tuple. If newest_alert_check_id() does not get exactly the same names as collect_alerts(), the alert's source differs, the acknowledgement stores check_id = None, _is_acknowledged() refuses to hide an alert without a check id, and the alert can never be closed. Keep the default () on build_alerts().
  • The closures that belong to a key are shown. app.closed_alerts() and database._grouped_closures() both filter on key_id IS NOT NULL, so a rotation, a replaced hardware, an updated contact, and an ignored key alert are listed, while remove_software and a software-sourced ignore are not. GET /api/health-checks/<id> returns them as closed_alerts (code, source, action, acknowledged_at), and the CSV has one last column, Follow-up closed, with code: action (YYYY-MM-DD). A closure hangs on the check it closed, through check_id.
  • The key card shows the dates and the note together. createKeyCard() in static/app.js always renders Last checked / Next check due, and adds the alert note on top. It used to replace the dates with the note, which cost the dates for a key whose only alert was the informational contact one.
  • The type is part of a key's identity (section 28). The setup locks it, the settings never change it, and only a rotation may pick the other one: POST /api/keys/<id>/rotate takes an optional type (hardware/cloud), and leaving it out keeps the current type. save_key_settings(..., key_type=None) uses COALESCE(?, type), so the settings and a rotation share one statement.
  • 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 409 through WalletSoftwareNameInUse — a ValidationError subclass, so it has to be caught before the general ValidationError in _rotate_key, or the 409 turns into a 400. validate_rotation() therefore takes the wallet software list as a fourth argument.
  • The settings page has two sections: 01 / Wallet software and 02 / Key rotation. The old 01 / Keys section is gone, because the name now belongs with the rest of the key identity.
  • 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 with Rotate this key and ticking the confirmation box (#rotation-replaced, "I confirm that each selected key has been replaced by a new key or hardware wallet") sends it through the rotate endpoint instead, together with the type and the new path. The single button is Save keys; it reports what was saved, and says so when there is nothing to save.

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 = 0 and a recorded address path that does not match. wallet_software and ai_security_checks must be empty; a non-empty value is a 400, not something dropped silently. Nothing is stored for them, and software_compatibility_status becomes NOT_ASSESSED.
  • Availability is required and offers only Yes/No. A cloud key collects only Availability; its compatibility is derived from it, and its own name is the wallet software.
  • A hardware key needs an explicit compatibility status, and Hardware works needs an explicit Yes or No. A missing value, "" and NOT_ASSESSED are all a 400.
  • 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_ASSESSED is not offered or accepted anywhere else. No field but the derived software_compatibility_status accepts it.
  • Address path verified is never a form field again. The server derives it from the comparison. The stored column stays, so the history, the export and the alerts keep working.
  • 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 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. rotate has its own endpoint, because it requires a new address path.
  • The export is plain CSV only. XLSX is not built and the app does not promise Excel compatibility.
  • The prohibition lists in sections 1 and 13 of TODO.md stay unticked. They are standing prohibitions, to be re-confirmed before release.
  • Whether a date is overdue is decided in the browser, so /api/key-status stays independent of the current date.
  • /api/previous-values still returns a security map that nothing consumes any more, and its address_path per key is no longer used by the form, which pre-fills the path from keys.address_path. Both are deliberate: removing them would shrink the endpoint, but nothing breaks while they stay.

Data model (schema version 11)

  • keys: key_number (1–3), name, type (hardware/cloud), address_path, created_at. There is no device column. 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 three rows, and the type only changes through a rotation.
  • health_checks: checked_at, next_check_due, wallet_software, software_version, software_compatibility_status, software_comments, contact_info_verified, comments, plus the legacy sparrow_version / nunchuk_version. wallet_software and contact_info_verified are free text.
  • key_checks: one row per submission — available, hardware_works, firmware_version, seed_backup_verified, address_path_verified, address_path, comments. The two cloud columns are gone.
  • ai_security_checks: provider (free text), status, comments, with UNIQUE (health_check_id, provider).
  • wallet_software: name, created_at. Stored whole, seeded with Sparrow Wallet when the list is empty, never emptied through the API.
  • alert_acknowledgements: code, source, key_id, check_id, reason, acknowledged_at. reason holds the action name that closed the alert, and key_id holds the key the alert belonged to, or NULL for a software alert from the settings list.
  • tool_checks no longer exists.
  • Startup migration rebuilds keys when it still has device, key_checks when it uses the old layout or still holds the two cloud columns, and health_checks and key_checks when their CHECK constraints still allow the old values. It then drops tool_checks if it is there, and adds the newer columns through OPTIONAL_COLUMNS — including alert_acknowledgements.key_id.
  • The table definitions are {table} templates, so a rebuild can create the same table under a temporary name. health_checks is a parent table, so the whole migration runs with PRAGMA foreign_keys = OFF, checks PRAGMA foreign_key_check before committing, and rolls back on any failure.
  • tests/test_database.py carries LEGACY_SCHEMA, SCHEMA_VERSION_5, SCHEMA_VERSION_8, SCHEMA_VERSION_9 and SCHEMA_VERSION_10 as literal older schemas. Copy that pattern for a new migration test: write the old schema, run initialize_database(), then assert on the migrated rows.
  • There is no data migration for an older test database: the user deletes the data folder and configures the setup again. An acknowledgement written before version 11 has no key, and the older contact alert used the source Contact 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. Endpoints: GET /health, 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.

The rotate payload is replaced + keep_name + name + type + address_path, and type is optional. GET /api/health-checks/<id> also returns closed_alerts. The settings page has two sections, and POST /api/settings/keys is what a rename alone goes through.

Five things that are easy to break:

  • New static files must be added to STATIC_FILES in app.py, or they return 404. That includes /settings and /settings.js.
  • /api/setup, /api/settings, /api/alerts, /api/key-status and /api/previous-values are extra endpoints that are not in the minimum list in section 11.
  • STORAGE_ERRORS in app.py includes RuntimeError, because list_keys() raises it for a partial setup. Keep it there, or a broken database answers with a traceback instead of a 500.
  • 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 a 404, a locked or duplicate address path is a 409, and a broken database is a 500. DuplicateAddressPath and WalletSoftwareNameInUse must be caught before the general ValidationError — in _save_key_settings and _rotate_key respectively — or the 409 turns into a 400.
  • newest_alert_check_id() in app.py derives 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 the 409 is built on, and _blocking_alert() reads the key id straight from the payload, so the rule applies before validation.

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 and TemporaryDirectory.cleanup() fails with WinError 32. Always close explicitly with try/finally; see write_schema() in tests/test_database.py.
  • A required control inside a display:none container 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 why renderSecurityChecks() shows an empty note instead of cards when a key could not be checked, why syncUncheckableState() clears the flags on the wallet software and the compatibility select — also for a cloud key, whose grid is hidden — and why syncRequiredFields() clears the flag on hardware_works and seed_backup_verified. The rotation fields follow the same rule: the type is disabled until the key is marked for rotation, the path is disabled when the key has a saved check and is not marked, and the script refuses to send an empty path.
  • The mismatch confirmation follows its warning. #address-path-confirmed is required only while #address-path-warning is visible, so the browser itself stops the submission and a required control is never hidden.
  • The page heading uses class="eyebrow" too. A regex over health-check.html finds Annual security check before the numbered sections. settings.html starts with Settings, then 01 / Wallet software and 02 / Key rotation. A test on the section order must expect that first. tests/test_health_check_form.py asserts those two lists exactly, so a new eyebrow on either page breaks it — the section 29 description line went on the dashboard only.
  • Do not assert on the quoting style of the JavaScript. static/health-check.js uses single quotes for constant selectors and template literals where a ${...} substitution is needed, and static/settings.js uses double quotes. Assert on the selector text, e.g. [name="address_path"], and on quote-free substrings such as /api/settings and /settings?rotate=.
  • clean_record() in tests/test_alerts.py has a fixed provider list and carries key_id and wallet_software. A test that needs another provider must replace the entry itself, and the recorded software is Sparrow Wallet, which is what a software alert names. A test that exercises the cloud-key form of software_incompatible has to pass cloud_names to build_alerts().
  • sample_payload() / unavailable_payload() are duplicated in tests/test_validation.py, tests/test_database.py and tests/test_app.py. They build the labels from the key's name and the selected software, they always send contact_info_verified, and for a hardware key hardware_works. They must send the key's own address_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.py also has ROTATION_PATHS and the rotate() helper, which picks an unused path per key so a test can rotate several keys in one run.
  • validate_health_check() takes a third argument (the wallet software list), and validate_rotation() takes a fourth for the same reason. tests/test_validation.py and tests/test_database.py wrap the first one in a small validate() helper.
  • tests/test_app.py has six test classes. ServerTestCase starts the server, AppHandlerTests stores the default setup in setUp, RenamedSetupTests stores the renamed one, SettingsApiTests, RotationApiTests and AcknowledgementApiTests store the default one, and KeySetupApiTests leaves the database empty. A new test must go in the class whose setup it needs.
  • Section 19 records one key per submission, so a Check id in the export can cover up to three keys only for rows saved before section 19. A legacy record with several key_checks is read through its first one when an alert needs the key it belongs to.
  • The database path default is repeated in sixteen places. database.py carries database_path="data/vardkey.sqlite3" as the default of every storage function (fifteen of them), and app.py carries it in get_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.