The instance's half of Halkyon's ADR 0044, step 5 (D37 to D40).
An invitation by mail (sharedSpace.inviteByEmail) goes out through the
register (send-space-invitation), naming who invites by their handle
and nothing of what is shared, with a link to /join/<token>. The
instance keeps only the address's keyed hash, as for sign-in, and the
SHA-256 of the token. Whoever opens it joins with a code mailed to that
address (EmailCodePurpose.joinSpace, proven through the register,
prove-address-code), so a forwarded mail admits nobody else; a code
proving another address is refused. It lapses after 14 days, and the
hourly pruning forgets it; inviting the address again replaces it.
A space's own link (setJoinLink, the Owner's) admits whoever opens it
as a Viewer (joinByLink) while it is on; switched off and on, it gets a
new token. Its token is kept as it is, so that whoever may invite copies
the link again (joinLink). The page at /join/<token>, in English and
German, says who invites and as whom; preview tells the app the same.
A membership someone else made is new to its member until they mark it
seen (isNew, addedBy; mine names who shared it), for the Inbox.
halkyon_app learns the two new refusals (handleNeeded, otherAddress),
which Learn reports as not allowed until its screens offer these ways
in. The ADR records D37 to D40 and how the instance's half is built.
The migration, the same in the module and in each server including it
(Learn, Meditate, Todo, Train):
ALTER TABLE "shared_space" ADD COLUMN "joinToken" text;
CREATE UNIQUE INDEX "shared_space_join_token_idx"
ON "shared_space" ("joinToken");
CREATE TABLE "shared_space_email_invitation" (…), with its unique
indexes on the space and address hash, and on the token's hash,
and its foreign key to "shared_space" ON DELETE CASCADE;
ALTER TABLE "shared_space_member" ADD COLUMN "addedBy" uuid;
ALTER TABLE "shared_space_member"
ADD COLUMN "isNew" boolean NOT NULL DEFAULT false;
Every space gets no token, and the unique index allows any number of
nulls; every membership gets no one who added it and isn't new, so
nothing already shared shows in an Inbox. The new table starts empty.
None of it rewrites a table, and the server before the switch names its
own columns, so it keeps working.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Codes by mail were tied to nothing but their address (the re-review's
seventh finding): a hostile instance could answer that a setup never
finished, have the person type a code into it, relay it to start the real
instance's account reset, and 72 hours later relay a second to finish it
with its own password and keys; the mail said "to sign in" whatever the
code was for. A code is now asked for a purpose (signUp, finishSetup,
authenticatorReset, accountReset, deletion), only that purpose takes it (any
other is refused as no longer valid, leaving the code as it was), and its
mail says what it would do ("to reset your account on learn.halkyon.eu: a
new password and new keys, and what was encrypted with the old ones is
lost") and to type it only in the app on that instance, nowhere else. The
app asks for each code with the purpose of its step.
Migration (module, Learn's and Todo's servers, Learn's device schema):
ALTER TABLE "email_code_request" ADD COLUMN "purpose" text; nullable,
so the requests of the last 15 minutes before it stay, taken by nothing
and forgotten within the hour.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Whoever knew the password could guess codes at 10 an hour forever,
unnoticed, each guess checked against every authenticator (the second
reading's M4). Every code an account's second factor refuses after a
right password (sign-in, a phrase's reset, a change) or from a signed-in
device (removing an authenticator, new recovery codes, the code from one
confirmed before) now counts, per account, from every network; after
every tenth in a row no code is taken (codesWait, saying until when), an
hour, then twice as long each time, a day at the most (longestCodeWait,
second_factor_waits.dart): at a day, ten guesses a day still have about a
1 % chance a year against one authenticator, so the first wait mails the
owner that someone knows their password. The instance keeps no address,
so sign-in and the phrase's reset now send the address the device typed
(and its language), used only for that mail after its hash is checked
against the account's. A code accepted clears it all.
Migration (module, Learn's and Todo's servers, Learn's device schema):
CREATE TABLE "second_factor_failures" ("id" bigserial PRIMARY KEY,
"authUserId" uuid NOT NULL, "failures" bigint NOT NULL, "waits"
bigint NOT NULL, "waitUntil" timestamp, "updatedAt" timestamp NOT
NULL) with a unique index on "authUserId": a new table, nothing
touched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The register's side of signing in with an address, a password that never
leaves the device (SRP-6a) and a code from an authenticator (ADR 0017),
with members forwarding every step through AccountRegister:
- password.set: salt, Argon2id's costs, SRP's verifier, the public keys
and the private keys wrapped twice, all made on the device; with the
address a mailed code proves, an anonymous person becomes identified
under their ID. It signs in once an authenticator is confirmed.
- passwordSignIn.begin and complete: SRP's two steps with a TOTP or
recovery code. An address nobody has is answered from a decoy key as
an account would be; every wrong credential is refused alike.
- authenticator: secrets sealed under totpKey (XChaCha20-Poly1305, bound
to their row), each code accepted once, ten recovery codes with the
first, the last one kept.
- accountRecovery: a lost authenticator reset 72 hours after the
password and a mailed code or the recovery phrase ask, a lost password
set anew with the phrase's signature, or the account reset with new
keys 72 hours after a mailed code asks; a mail at each start, a
sign-in calling it off.
- An account with a password no longer signs in by a mailed code or
GitHub; no new GitHub account is linked where passwords are offered,
and from GITHUB_SIGN_IN_ENDS GitHub signs nobody in and its IDs go.
Five tables, added by the module's migration; deletePerson deletes
their rows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
