DodoTech-Public/DodoSSH
Public Access
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 2 Wiki Activity

Merge branch 'main' into the vaults screen, and let it rotate keys too

Browse Source 
Main built vault key rotation while this branch was reshaping the screen that
would drive it, so the two met in the same three files. Every other conflict was
textual and resolved by taking both; these are the ones where a decision had to
be made.

**The view model.** Main taught TeamsViewModel three things and this branch had
renamed and rewritten it into VaultsViewModel. All three are ported rather than
dropped, because each is a behaviour rather than wording: adding somebody now
wraps the vault to them on the spot instead of leaving SHARE KEY to be pressed,
removing somebody rotates the vault and hands the new key to whoever is left, and
a share reports how many generations were wrapped. The session calls they reach —
ShareTeamVaultsAsync and RekeyTeamVaultsAsync — are scoped to a membership list
rather than to one vault, and they are called that way here rather than narrowed:
adding somebody is a change to the list, so every vault the list carries is one
they can now fetch. This screen makes lists that carry one vault, so the sentences
name one; where a list carries several, naming them all is the honest report, and
the members section already says the list is shared.

AddMemberAsync ran two lines over the length limit once the sharing was in it, so
the calls behind it moved to AddOrInviteAsync and the three-way refusal to
WhyNobodyCanBeAdded — the command reads as its guards now, which is what it was
before the sharing arrived.

**The tests.** Main's four new cases are ported to the vault-first API, including
the one that matters most: the tampered key log is corrupted *before* the add,
because the add is now a route to a wrap and a test that corrupted it afterwards
would be asserting about the manual route only. SelectingAVault_ListsWhoHoldsAKey
now expects two holders rather than one — main's fake records the creator's own
self-grant, and a key-holder list that omitted it would show the one person who
can certainly open a new vault as somebody who cannot.

**The README.** The limits list is six rather than four or five: main's rotation
entries and this branch's "a vault cannot be deleted" describe different things
and both are true. "The rekey is flagged, never performed" is gone, since it is
now performed, and M3 reads *Done* rather than *Done, except rekey*.

One thing worth writing down that neither side had. An invitation claimed at
sign-in still leaves the key owed, where an add does not: at the moment an
invitation is issued there is no account and no published key to wrap to, and the
claim happens on the invitee's machine, which holds nothing. Manual check 12.1
says so, because a reader who knows adding shares would otherwise read that step
as stale.

1561 tests pass.
This commit is contained in:
jaap-jan
2026-08-04 13:58:56 +02:00
parent 8707629a6c 8d2f4c8ffb
commit a0568d4c35
51 changed files with 4785 additions and 275 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/adr/0009-team-access-model.md
+6
View File
@@ -67,6 +67,12 @@ One thing is deliberately **not** built, and it is a refusal rather than an omis
- **The rekey itself.** Only a client holding the current vault key can re-wrap every item's data key
under a new one. The server records that a rotation is owed and the interface reports it. M5.
> **Superseded 2026-08-03 by [ADR 0010](0010-vault-key-rotation.md).** Rotation now ships, and it
> turned out to divide differently than this paragraph assumed: advancing the generation is one
> server transaction and is not the same act as re-wrapping the items, which is still outstanding.
> Removing a member rotates the vaults the removing client can open and hands the new key to whoever
> is left.
Two smaller choices, recorded because the alternative was written down first and rejected:
- **No `v_user_vault_permission` view.** ADR-adjacent notes and the old `VaultAccessService` remark
docs/adr/0010-vault-key-rotation.md
+144
View File
@@ -0,0 +1,144 @@
# ADR 0010 — Rotation advances a generation; the keys before it are kept
- Status: accepted
- Date: 2026-08-03
- Builds on: [ADR 0001](0001-e2ee-trust-model.md), [ADR 0009](0009-team-access-model.md)
## Context
ADR 0009 shipped removal as "revoke the grants and flag the vault", and named the missing half
plainly: only a client holding the current vault key can produce the next one, so the server could
record that a rotation was owed and nothing more. Nothing acted on the flag. In practice that meant
removing somebody from a team left every vault they could read encrypted under the key they had, for
ever — the interface said a rotation was owed and no button existed to perform one.
Two things had to be decided before that flag could be acted on, and they are not independent.
**When does the generation change?** A vault key is per vault *per generation* (`docs/crypto.md` §3),
and a grant names the generation it opens. If two admins rotate at the same time, both wrap a key,
both issue grants, and the vault ends up with two claimed "current" keys and a set of members split
between them — half of whom cannot read what the other half writes, with nothing to point at as the
cause.
**What happens to everything already stored?** An item carries the generation it was sealed under, in
its own row and in its AAD. A rotation that advanced the generation and left the old grants behind
would make every item written before it unreadable to everybody, including the person who rotated. A
rotation that re-encrypted every item would avoid that — and is a different, much larger operation:
`crypto.md` §3 puts it at N × 32 bytes of re-wrapped data keys, which is cheap in bytes and is still
a write to every row of a vault, in batches, against a server that caps a push at 500 operations and
8 MB, with the connection/activity logs alone reaching five thousand entries per kind.
## Decision
**The rotation is the generation bump, and it is one server transaction. Grants for earlier
generations are kept.**
`POST /api/v1/vaults/{id}/rekey` takes the next generation and the new key sealed to the caller. In
one transaction the vault's `key_generation` advances, the caller's grant for it is inserted, and the
rekey flag is cleared. The request must name exactly `current + 1`, and the vault's `xmin`
concurrency token makes that check binding rather than advisory — the second of two simultaneous
rotations is refused and told to read the vault again. The server contributes the *moment*, which is
the one part of a rotation a client cannot decide for itself; it contributes no cryptography, cannot
tell that the key it is handed differs from the old one, and cannot tell whether the caller held the
old one. That last part is checked the only way it can be: the caller must hold a live grant at the
current generation, which is a row rather than a proof.
Everything that follows from keeping the old grants:
- **A member holds one grant per generation, and `VaultSummary` serves all of them.** The current
wrap stays where it was; the rest arrive as `PriorKeyWraps`, oldest first. `VaultKeyring` holds a
key per generation, hands out the newest for writing and the item's own for reading. Every read
path picks its key from the payload's `keyGeneration` rather than from the vault's.
- **Sharing hands over the history.** `ShareVaultAsync` issues a grant for every generation the
sharing client holds, oldest first. Somebody added after a rotation who was given only the newest
key would open the vault to a list of items that will not decrypt — which reads as corruption, not
as a missing grant. The server accordingly accepts a grant for any generation the vault has
reached, and refuses one for a generation ahead of it: nothing is sealed under that, and accepting
it would let a client move the vault forward outside the transaction that is allowed to.
- **Revocation takes every generation.** Removing a member, and `RevokeGrantAsync`, revoke all of a
recipient's grants rather than the current one. Leaving the history would leave them able to read
everything written before the rotation, which is exactly what the rotation was for.
- **A member between the rotation and their re-wrap can read and cannot write.** They hold the
history and no current key, so the vault lists as unreadable and writes refuse. Writing under a
superseded key would produce items nobody else could open, and the author's own keyring — which
still holds that key — would show no sign of it.
**Removing a member rotates automatically.** The teams screen removes the member, then rotates every
team vault the machine can currently open and wraps each new key to the members who remain. Adding a
member is the mirror image: every team vault this machine can open is wrapped to them as part of the
add. Both report per vault, including what they could not do — a vault whose key this machine does
not hold is skipped and stays flagged, because somebody else has to finish it.
## The second half: re-sealing what is already stored
> **Added 2026-08-04.** This was deferred when the decision above was taken, and is now built. The
> reasoning that made it safe to defer is what made it cheap to add, so it is recorded here rather
> than in an ADR of its own.
A rotation on its own re-keys the vault and not its contents, which leaves one gap: somebody who left
with a copy of the old key could still open old ciphertext they later got hold of. `VaultResealer`
closes it by walking the vault and rewriting each item under the current key, as an ordinary upsert
against the version the server holds.
Four properties, each of which is a decision:
- **It never decodes the plaintext.** An item is opened and the *same bytes* are sealed again under a
fresh data key. No codec, no merge, no schema version — so an item written by a newer client
survives untouched, where re-encoding it through this build's codec would silently drop the fields
this build has no concept of. It is also why one pass covers every item type, including types added
after it was written.
- **It is resumable, and needs no transaction.** Each item is one upsert, so a pass that dies half way
leaves a vault at mixed generations — which is a state that reads perfectly well, because that is
precisely what the decision above bought. Running it again picks up what is left.
- **A conflict is counted, not merged.** The pass changes no content, so there is nothing to merge:
an item somebody else wrote meanwhile is left at their version and re-sealed on the next pass.
- **A queued local edit is left alone, and re-sealed on the way out instead.** Rewriting it here would
overwrite the user's unpushed work with the version the server holds. Instead `SyncEngine` re-seals
a queued payload whose generation is stale as it dispatches it, and writes the revision back to the
outbox first so a retry sends the same bytes. That closes the one hole a pass over *stored* items
cannot see: a change made before the rotation and pushed after it would otherwise put a brand-new
item into the vault under the key the departed member holds.
The pass runs as the last step of a rotation, after a sync — a mirror that is behind produces a batch
of conflicts rather than a re-sealed vault. The interface reports which of the two guarantees was
reached, because they are different: a vault fully re-sealed is closed to the person who left, and one
where items were left behind is closed only to what happens next.
## What this still does not do
**It does not reach what they already pulled.** The person who left keeps whatever plaintext is on
their machine — that is the non-retroactive limit ADR 0001 records and no design here changes it. The
honest remediation for a departure is still to rotate the credentials themselves, and the product says
so rather than the reassuring version.
## Alternatives rejected
- **Revoke the old grants on rotation.** Tidier, and it makes the grant list say exactly one thing
per member. It also makes every item written before the rotation unreadable to everybody, which is
data loss performed by a security feature.
- **Chain the keys: store each old key sealed under its successor.** One wrap per rotation instead of
one grant per member per generation, and new members get the history for free. It needs a new table,
a new AAD purpose, and a recursive unwrap on the read path — and it makes the vault's whole history
reachable from the current key, which is a strictly larger blast radius than a set of grants that
can be revoked one at a time.
- **Rotate atomically with every item re-sealed, in one request.** The safest shape on paper and the
one `crypto.md` implies. It caps rotation at the push limits — 500 operations and 8 MB — which a
vault with a year of connection log in it exceeds, and the failure mode is a vault that can never
be rotated at all.
- **Let the server generate the new key.** It would make rotation a single call and would end the
product: a server that can produce a vault key can read the vault.
## Consequences
`vault_key_grant` grows by one row per member per rotation. The unique index is already per
`(vault, generation, recipient)`, so this needed no migration; the rows are 110-byte seals and a vault
rotated monthly for a decade with ten members holds twelve hundred of them.
The sharing graph gains a dimension the operator can read: which generation each member holds, and so
which of them have been re-wrapped since the last rotation. That is the same class of metadata ADR
0009 already records as visible, and it is the same fact the sharing screen shows the members
themselves.
A client that never comes back holds keys to generations that no longer receive writes, which is the
same exposure as any copy of a vault key on a machine that has been lost — bounded by the fact that
the server will not serve them anything, and unbounded in the way every non-retroactive revocation is.
docs/adr/0011-android-distribution.md
+125
View File
@@ -0,0 +1,125 @@
# ADR 0011 — Distributing the Android client, and who holds the release key
- Status: accepted
- Date: 2026-08-03
- Builds on: [ADR 0001](0001-e2ee-trust-model.md)
- Settles: the second open question in [`docs/android-port.md`](../android-port.md#still-open)
## Context
[ADR 0001](0001-e2ee-trust-model.md) ends on the hole it cannot close with cryptography: **an operator
who wants the secrets attacks the client, not the crypto**, and what that costs is *release signing with
a key not held by the server*, and eventually reproducible builds. Until now that sentence had nothing
to bind to. The desktop head is not packaged at all — packaging is M4 — and the Android head is a debug
artefact: CI runs `-t:SignAndroidPackage` with no keystore, so it is signed with the debug key .NET for
Android falls back to when `AndroidKeyStore` is false, and that APK is a build check rather than
something anyone installs.
The first release changes that, and it does so **irreversibly**, which is why this is decided here
rather than at upload time. Two Android facts make it a one-way door:
- **An installed app can only be updated by a package signed with the same key.** The signing key is
the app's identity for its whole life; changing it means every existing user uninstalls first, losing
their local cache and re-enrolling. (v3 signature rotation exists, but the lineage has to be created
*before* it is needed, by the key it is rotating away from.)
- **A new app on Google Play must ship as an App Bundle, which means Play App Signing**, so Google
generates and holds the key that signs what users install; the developer holds an upload key only.
There is no un-enrolling. For an app that already exists outside Play, the only way to keep one
package id across both channels is to *hand Google the existing key*.
So "publish on Play" and "hold our own key" are not two settings. They are two package identities, and
the first release picks one.
The third party changes but the shape does not: **whoever can sign an update can ship one person a
build that copies the passphrase.** The vault's encryption is irrelevant to that attack — the client is
where the plaintext is, by construction (ADR 0001, `Connect` cannot be a security boundary). So this ADR
is about *which* parties are in that position, not about removing them, and there are three candidates:
the deployment operator, DodoTech, and Google.
## Decision
**DodoTech holds the release key, the deployment never serves the client, and Play is a separate
decision that has not been taken.**
1. **One release key, held by the project, kept offline, and never in CI.** Release signing is a
deliberate manual step on a machine that is not a runner. CI keeps doing exactly what it does now —
packaging with the debug key to catch link-time failures — and must never gain a keystore secret or
an `AndroidKeyStore=true`. A signing key in CI is a key held by whoever can push a workflow file,
which for a public repository is a wider set than it looks.
2. **The APK is published on the project's own release page, and a DodoSSH deployment never distributes
it.** This is the refusal that carries the security content, and it is the one a self-hosted product
gets wrong by default: a "download the app" link on your own server is convenient, obvious, and hands
the client binary to the exact party ADR 0001 models as the adversary. The operator may tell people
where to get it. They may not be the place it comes from.
The same rule reaches the update path. A version check pointed at the deployment lets the operator
pin a chosen user to a known-vulnerable build by withholding the answer — a weaker attack than
signing one, and available without any key at all. If an update check is ever added it points at the
project's domain, and the first release simply has none: the release page is the channel and the
README says so.
3. **Play App Signing is not entered, and cannot be entered by accident.** Not because Google is a worse
custodian than DodoTech — on the mechanics it is a better one, since the key lives in Google's
infrastructure rather than on a laptop, and the reach and auto-update story is not close. It is
declined because of what it costs *this* product specifically: the buyers named in ADR 0001 are teams
who refuse to put infrastructure credentials in a SaaS, and telling them the client that holds their
plaintext is signed by a key the vendor cannot see is the same answer they already rejected, one
layer down. A targeted signed build compelled by a lawful order or produced from a compromised
console account is the archetype of the attack ADR 0001 calls the largest practical hole.
Deferring is cheap and reversing is not, so the default falls the deferrable way. Revisiting is a
second ADR, and it has two honest exits: hand Google the existing key and keep one identity, or take
a distinct package id and accept two apps. Both are worse decisions to discover than to take.
4. **Reproducible builds are the goal that makes all of the above matter less, and they are not
achievable today.** A build a third party can reproduce from source turns the signing key from a
trusted authority into a convenience — anyone can check that the published APK is the published
source. .NET for Android is not there: dex output, AOT images and archive timestamps are not
bit-reproducible across machines in practice. It stays the standing goal ADR 0001 names, recorded
here as the thing that would let point 3 be reconsidered on the merits rather than on custody.
5. **F-Droid is not a channel.** Its build server compiles from source and signs with its own key, which
would be a genuinely better transparency story — but it has no support for a .NET workload plus an
Android SDK toolchain, and this head needs both. Not refused; unavailable.
## Consequences
**The reach cost is real and should not be talked down.** Installing means enabling installation from
the browser or file manager, per source, on Android 8 and later — a permission the platform frames as
dangerous, correctly. There is no discovery, no automatic update, and no Play channel for a corporate
MDM to deploy from, which for a product sold to teams is the sharpest edge of this decision. What
partially answers it is that this client is installed by people who already run their own identity
provider and their own vault server; sideloading is not the strangest thing they will do that week.
**The key becomes a single point of failure with no recovery.** Losing it means existing installs can
never be updated again — not a bad update, *no* update — and the only way out is a new package id and a
manual migration. It is therefore backed up offline in more than one place, and a v3 rotation lineage is
created at the first release rather than at the first emergency, because a lineage can only be signed by
the key it replaces.
**The attack ADR 0001 names is narrowed, not removed.** DodoTech can still ship one user a malicious
build. What changes is that the deployment operator — the party the threat model is actually about, and
the one with a motive to read their own team's credentials — cannot, and that a compelled or breached
third-party store is not in the path either. That is the whole of what this decision buys, and it is
worth stating at that size rather than larger.
**M4's desktop packaging inherits rule 2 and not the rest.** Windows and macOS have no equivalent of
Play App Signing in the mandatory sense: Authenticode and Developer ID both leave the private key with
the developer, and Apple's notarization is a scan rather than a signature over the shipped binary. The
custody question is therefore easy there; the "not served by the deployment" rule is the part that
carries over, and it carries over unchanged.
## Rejected
- **Play as the primary channel, sideloading as the fallback.** This is the arrangement most Android
products land on, and it does not survive contact with rule 2's reasoning: it is the same shape —
the binary that holds the plaintext arriving through a party who can be compelled — with a larger and
better-resourced party in the middle. Better mechanics, same class.
- **Shipping the APK from the DodoSSH server it will talk to**, so a new phone gets the client from the
deployment it is enrolling against. Genuinely the nicest onboarding available, and it makes the
operator the distributor of the client that holds their team's credentials. Refused outright, and
named here because it will be proposed again.
- **A release key held by CI so tagging cuts a release.** The convenience is the point of CI and the
key is the point of this ADR; where they collide the key wins. Signing an artefact is one command, run
rarely, by a person.
docs/android-port.md
+44 -13
View File
@@ -28,7 +28,9 @@ verified is that it compiles, links, packages, and carries the right natives.
**The shape agreed:** a **phone-first** client that is the keychain plus a **terminal**, with sessions and
transfers protected by a **foreground service**. File transfer is not in the first scope; when it arrives it
is **one remote pane** with Android's document picker for moving files in and out.
is **one remote pane** with Android's document picker for moving files in and out. *It has since arrived,
both ways:* the pane, the queue, `ACTION_OPEN_DOCUMENT` going in and `ACTION_CREATE_DOCUMENT` coming out,
with the foreground service now counting transfers as well as shells.
**What was actually checked**, so the rest can be read with the right amount of trust:
@@ -132,7 +134,9 @@ mean the transfer queue's local half (`LocalDirectory`, the drive list, the brea
code.
Note what *does* carry: `FileTransferQueue` itself, and `IRemoteFileStore` — Phase 6 already proved that
seam holds two very different remotes, and a `Uri`-backed Android document would be a third.
seam holds two very different remotes, and a `Uri`-backed Android document would be a third. (**⚠️ That last
clause is wrong, and the build is what corrected it**: a document URI cannot resume, so what shipped stages a
copy and hands over a path. See the note under the decision below.)
**Decided: one remote pane and the document picker**, and out of the first scope. See
[Decisions](#file-transfer-when-it-comes-one-pane-and-the-document-picker).
@@ -256,6 +260,24 @@ and `IRemoteFileStore` both carry over unchanged. Phase 6 already put a bucket b
an SFTP host, so a picker-granted document is a third implementation of a seam that has been exercised twice.
What is desktop-only is the *left* pane — `LocalDirectory`, the drive list, the breadcrumb trail.
> **⚠️ Corrected by the build. Both directions shipped, and neither as a third `IRemoteFileStore`.** A document
> URI cannot sit behind that interface honestly: the queue seeks, because an upload resumes from the byte the
> last attempt reached, and a `content://` stream promises no seek, no stable length, and no grant that
> survives the document being edited underneath it. So `DocumentStaging` copies the chosen document into the
> application's cache and hands the queue an ordinary path — which cost one class in the head and *nothing*
> in the shared layers, rather than a third implementation of a seam and every resume rule rewritten to cope
> with a stream that cannot rewind. The copy is deleted when the transfer completes, kept while it is stopped
> so RESUME can read it, and swept at the next launch.
>
> **Outbound is the mirror image, with one decision the inbound half did not have to take: when to ask.** The
> save picker is raised *before* the transfer, so `QueueDeliveredDownload` runs into the same staging
> directory and hands the finished bytes to a callback the head supplied. Asking afterwards would put the
> picker minutes away from the button that caused it and, on a phone, frequently in the background — where
> Android will not show one. The cost is that `ACTION_CREATE_DOCUMENT` creates its file when it is
> dismissed, so a download that then fails leaves an empty one; that is said on the screen and in the
> README rather than left to be found. A delivery that fails keeps the staged bytes for the sweep instead of
> deleting the one copy of something that was just fetched over somebody's network.
### Sessions survive backgrounding, via a foreground service
A persistent notification for as long as a shell or a transfer is live.
@@ -577,9 +599,11 @@ What is left, in the order it matters:
- **Running any of it on a device.** Still the one that is not optional, and still true: nothing here has
ever been launched on hardware or an emulator. Everything below is reasoning from the code.
- **The document picker.** The files screen browses a remote, downloads from it and runs the queue; putting
a file *on* a host from the phone needs `ACTION_OPEN_DOCUMENT` behind a third `IRemoteFileStore`-shaped
source, which is the decision recorded above and the obvious next piece of work.
- ~~**The save picker — the other half of file transfer.**~~ **Built**, along with the half before it: both
pickers go through Avalonia's storage provider and both stage through the cache. See the correction under
the decision above for the shape and for the one thing about it a person will notice — an empty file where
a failed download was pointed. What is *not* built is a folder picker for several downloads at once: the
save picker names one destination, so SAVE FILE takes the selected row.
- **Editors.** There is no host editor and no keychain item editor on the phone, so both are create-on-
desktop-and-sync. That is why the v2 design's `+` buttons on HOSTS and on the keychain are not drawn.
- **Pins and import**, which v2 does not draw either. Teams *is* drawn, behind MORE — it was the one of
@@ -590,7 +614,9 @@ What is left, in the order it matters:
## Still open
Neither of these blocks the spike, and both want answering before there is anything to release.
Neither of these blocked the spike, and both wanted answering before there was anything to release. Both
now have answers, kept here under the questions that produced them — the heading stays because what is
worth reading is which way each went and why, not that the list is empty.
- ~~**Which Android versions.**~~ **✅ Settled: minSdk 28, targetSdk 36**, and the reasoning divided in two.
@@ -606,13 +632,18 @@ Neither of these blocks the spike, and both want answering before there is anyth
devices that mostly cannot hold a hardware-backed key anyway — which is the one thing the store is for.
API 28 and 29 still cost one branch each in `BiometricGate`, because allowed-authenticator lists arrived
in 30.
- **How it is distributed, and what that does to the supply-chain story.** ADR 0001 says plainly that an
operator who wants the secrets attacks the client rather than the crypto, and that release signing with a
key **not held by the server** is what that costs. Play App Signing means Google holds the release key.
That is not necessarily wrong — it is a different, and in some ways better-audited, trust arrangement —
but it is a change to a documented security property of this product, and it should be reasoned about in
an ADR rather than discovered at upload time. Sideloading a self-signed APK preserves the current story and
costs reach.
- ~~**How it is distributed, and what that does to the supply-chain story.**~~ **✅ Settled in
[ADR 0011](adr/0011-android-distribution.md): the project holds the release key, the deployment never
serves the client, and Play is deferred.** The question was whether Play App Signing — Google generating
and holding the release key — is a change to the security property ADR 0001 documents. It is, and the ADR
takes it as a one-way door rather than a setting: a new Play app must use App Bundles and therefore Play
App Signing, an installed app can only be updated by a package signed with the same key, so the first
release picks one identity for good.
Two things follow for this head *now*, before there is anything to release. CI keeps signing with the
debug key and must never gain a keystore secret. And no download link for the APK may ever be served by a
DodoSSH deployment — that hands the client binary to the party the threat model is about, which is a
worse arrangement than either of the two the question was originally between.
## Smaller things, decided by default
docs/crypto.md
+16 -2
View File
@@ -216,13 +216,27 @@ This is the load-bearing structural choice. Because every wrap protects the *sam
### Why a per-item DataKey
1. **Cheap rotation.** Rotating a vault key re-wraps N × 32-byte data keys and never touches
content blobs. A 10,000-item vault rotates in a few hundred kilobytes of writes.
1. **Cheap rotation.** Re-keying N items re-wraps N × 32-byte data keys and never touches
content blobs. A 10,000-item vault re-keys in a few hundred kilobytes of writes.
2. **Narrow sharing.** A single item can be re-wrapped to another vault key or user key.
3. **Nonce hygiene.** Each key encrypts about one message.
4. **Versioning.** A new item version gets a new data key, so prior ciphertext stays
independently decryptable for history and undo.
> **Added 2026-08-03: what a vault key rotation actually does.** Advancing a vault to a new
> generation does **not** re-wrap the items already in it. Each item keeps the generation it was
> sealed under, in its row and in its AAD, so a rotated vault holds items under two or three keys
> at once and every read chooses the key its item names. That is why a member's grants for earlier
> generations are kept rather than revoked, why `VaultSummary` serves all of them, and why sharing
> issues one grant per generation held: a client holding only the newest key would read the vault's
> whole history as tag failures.
>
> Moving those items onto the new key is a **separate pass** (`VaultResealer`), run last and resumable
> because a vault at mixed generations is readable. It opens each item and seals the same plaintext
> again under a fresh data key — the envelope is re-made rather than only the wrap, because the AAD
> binds `keyGeneration` into both. The plaintext is never decoded, so an item written by a newer client
> crosses a rotation untouched. See [ADR 0010](adr/0010-vault-key-rotation.md).
Per-item keys wrapped *to individual users* — which is what would make per-item ACLs
cryptographic rather than server-enforced — are deferred to M5. The `content_key_id` column
exists from the first migration so that lands without a migration. Until then, **an item ACL
docs/design-import-gaps.md
+6 -4
View File
@@ -46,7 +46,7 @@ the chrome, hosts and terminals, file transfer, the vault, teams, and preference
> | Logs: a blinking **FOLLOW** pill over a live `tail` | A REFRESH button. These are the keychain's own synced audit records, written once at close and read when the screen opens — there is nothing to tail. |
> | Logs: `ALL 8 / INFO 5 / WARN 2 / ERROR 1` and per-host chips | The real two-way choice: connections, or keychain activity. A connection record is not a syslog line and carries no level, and neither log filters by host. |
> | SFTP: the `LOCAL / REMOTE` toggle | One remote pane. Scoped storage means there is no browsable local filesystem, which `docs/android-port.md` decided before any of this was built. |
> | SFTP: the transfer card, and moving files at all | ◆ **Browsing, opening and deleting only, and the screen says so.** Both transfer commands exist and both work — against the local pane. `QueueDownloads` writes to `Path.Combine(LocalPath, name)`, and `LocalPath` starts at `SpecialFolder.UserProfile`, which on Android is the application's own private directory: a download would report success and leave the file somewhere the person who asked for it cannot open. That is worse than not offering it, because a refusal is visible and a file in `/data/user/0/…` is not. The queue is not drawn either, since nothing on this head can put anything in it. Both return with the document picker. |
> | SFTP: the transfer card, and moving files at all | ◆ **Both directions ship, and neither is the button the design drew.** There is no local pane for an UPLOAD arrow to point away from, so the gestures are the platform's two pickers: ADD FILES is `ACTION_OPEN_DOCUMENT` — "point at a document wherever it lives" — and SAVE FILE is `ACTION_CREATE_DOCUMENT`, asked *before* the transfer because a picker raised on completion arrives over whatever the person moved on to, and often while the app is backgrounded and Android will not show one. Both stage through the cache, for the same reason: a `content://` URI has no path, no trustworthy length and no promised seek, and the queue needs all three to resume. `QueueDownloads` — the desktop's path, writing into `LocalPath` — is unused here, because `LocalPath` on Android is the app's own private directory. The queue is drawn now, since this head can fill it. |
> | S3: `96 objects · 412 GB` and `lifecycle: IA after 7d` | Omitted. Neither a bucket total nor a lifecycle policy is fetched; the listing is what `IRemoteFileStore` returns. |
> | A `+` floating button on **HOSTS** | **Shipped in v3.** It was omitted twice, on the grounds that it would open a host editor this head had not got — and that was true until one existed. It raises a sheet rather than a form, because "add" on this screen is two operations: a host, or a group to file hosts under. |
> | A 14px rounded inset around the terminal | Not drawn. The renderer is a native child view composited above everything Avalonia draws, so a rounded frame behind it clips nothing. What the design was after comes from the page's own background, which is the design's `#171A26`. |
@@ -429,9 +429,12 @@ answerable by anybody willing to create a team first. It simply gets claimed soo
| Shared vaults | server + client | A team owns vaults; each is created with the creator's own grant, because a vault with no grant is a container nobody can open. |
| Roles | contracts + server | `TeamMemberRole` on the wire, numerically pinned to `DodoSSH.Domain.TeamRole` by a test. Viewer reads, Member writes, Admin and Owner also share and administer. |
| Members table | server | `TeamMemberSummary`, and a directory that resolves an exact email to a public key. |
| Sharing an item | client | `VaultSession.ShareVaultAsync`: verify the recipient's key against the key log, wrap, sign, record. The server stores the wrap and the signature and can check neither. |
| Sharing an item | client | `VaultSession.ShareVaultAsync`: verify the recipient's key against the key log, wrap, sign, record. The server stores the wrap and the signature and can check neither. One grant per generation the sharing client holds, so a recipient can read a rotated vault's history and not only what happens next. |
| Adding a member shares the team's vaults | client | Adding somebody wraps every team vault the adding machine can open to them, as part of the add rather than as a button to remember. Membership and a key are still two acts on two machines; the client just performs both. A vault this machine holds no key to is skipped and named. |
| Removing a member rotates the vaults | server + client | `POST /api/v1/vaults/{id}/rekey` advances the generation and records the caller's new grant in one transaction — the server contributes the moment and no cryptography. The client then wraps the new key to the members who remain. Grants for earlier generations are kept, or the vault's stored items would become unreadable to everybody. See [ADR 0010](adr/0010-vault-key-rotation.md). |
| Re-sealing a rotated vault's stored items | client | `VaultResealer`, the last step of a rotation: every item is opened and the same plaintext sealed again under the new key, in batches, against the version the server holds. The plaintext is never decoded, so an item written by a newer client crosses a rotation untouched. Resumable, because a vault at mixed generations is readable — a pass that stops half way leaves a working vault and is re-run. A change queued before the rotation is re-sealed by the push path instead, so nothing reaches the server under a superseded key at all. |
| Pending invites, and withdrawing one | server | A `team_invitation` row per (team, address), listed beside the members it is about and withdrawable until it is taken up. It becomes a membership when an account with that address signs in — **and only if the access token asserts `email_verified`**, because membership is authorisation and an invitation anybody could take by naming somebody else's address is a way in. Fourteen days, because an address that is reassigned would otherwise carry a standing offer to whoever holds the job next. |
| Ownership transfer | server | `POST /api/v1/teams/{id}/owner`, owner only. One transaction: the named member becomes owner and the outgoing owner becomes an admin. Not two role changes — ownership is sole, so promoting first leaves the team owned twice and demoting first leaves it owned by nobody. The outgoing owner is demoted rather than removed, because removing them would revoke their vault key grants and flag every team vault for rekey, which is a far larger act than the one being asked for. |
| Ownership transfer | server | `POST /api/v1/teams/{id}/owner`, owner only. One transaction: the named member becomes owner and the outgoing owner becomes an admin. Not two role changes — ownership is sole, so promoting first leaves the team owned twice and demoting first leaves it owned by nobody. The outgoing owner is demoted rather than removed, because removing them would revoke their vault key grants and rotate every team vault, which is a far larger act than the one being asked for. |
| `LAST ACTIVE` | server | Real, and coarse on purpose. `UserAccount.LastSeenAtUtc` is now refreshed on ordinary authenticated requests, at most once per account per hour: writing it per request would put an UPDATE on the hot path of every authenticated call and start losing races on `user_account`'s own concurrency token. So the column answers "this week or not", which is the granularity the question is actually asked at, and is shown coarsely rather than to the minute. |
| Renaming and archiving a team | server | `PUT` and `DELETE /api/v1/teams/{id}`. The slug is deliberately not renameable: it is unique only among *live* teams, so a rename could take a slug an archived team still holds and strand it. Archiving soft-deletes the team, every membership and every pending invitation in one transaction — and is refused outright while the team owns any vault. |
@@ -443,7 +446,6 @@ answerable by anybody willing to create a team first. It simply gets claimed soo
| The invitation mail, and **resend** | server | An outbound mail path: an SMTP configuration, a template, a bounce story and a deliverability problem, none of which this server has. | **Nothing is sent, and the interface says so.** An invitation is a standing instruction rather than a message — the next account to sign in with that address joins the team — so there is no token, no link, and nothing to resend. Telling somebody to sign in is done over a channel this server does not carry. A link nobody can deliver would be worse than no link. |
| Archiving a team that owns vaults | — | Nothing that would be safe. A team vault resolves through membership, so archiving would take those vaults away from everybody holding a key, silently, including the caller — and nothing in this product deletes a vault, so there is no sequence of calls that turns the refusal into a success. | Refused, with `team-not-empty` and a count of the vaults in the way. A stated limit rather than a coming feature, for the reason the SFTP layer refuses a recursive delete: a refusal is visible and a quiet removal is not. |
| `SSO · OIDC · okta.dodotech.dev` | server | Per-team SSO. Authentication is one global JWT scheme bound to one authority. | Omitted. |
| A rekey after a membership change | client | Re-wrapping every item's data key under a fresh vault key, which only a client holding the current one can do. M5. | The vault is flagged `RekeyRequired` and the row says a rotation is owed. |
> **The trap this document warned about is still a trap.** `GET /api/v1/meta` advertises
> `features: ["teams"]` *unconditionally* (`MetaEndpoints.cs`). It was meaningless when nothing implemented
docs/manual-checks.md
+124 -1
View File
@@ -1158,7 +1158,11 @@ follows is about what happens the first time it does.
and Bob is now in the members table.
6. **Pass, and this is the half that is easiest to lose:** the vault is in Bob's list **saying it is
waiting for a key**, and nothing in it is readable. Have Alice press SHARE KEY and Bob sync; now it
opens.
opens. This is where an invitation differs from an add and the difference is not a defect: adding an
account the directory already knows wraps the key on the spot, because the adding machine has it and
the recipient has a published key to wrap to. An invitation has neither at the moment it is issued —
there is no account yet — and the claim happens on Bob's machine, which holds nothing. So the key is
still owed, and somebody has to hand it over.
**Failure means:** step 5 failing with everything else passing is almost always the `email_verified` claim
— go to 12.2 rather than reading the invitation code, because the server is doing exactly what it should.
@@ -1342,3 +1346,122 @@ rather than two. There is no confirmation prompt, deliberately.
**Failure means:** a phone that still unlocks itself after this is the local half not happening, which is the
half that matters when the handset is the thing that was lost.
---
## Phase 14 — Moving files to and from the phone's remote
Every check here needs a real Android device or emulator, a host with SFTP or a bucket, and at least one
document on the phone to send. What is automated is what can be: `TransferQueueingTests` says a staged
upload obeys the same rules as any other and that a delivered download refuses a directory before the
picker's damage is done. What cannot be automated is everything below — the two system pickers are another
application, and the staging copies, the delivery, the notification and the resume are all things only a
running phone does.
### 14.1 ADD FILES opens the system picker, and takes more than one
Connect to a host on SFTP, navigate somewhere writable, press **ADD FILES**, and choose two documents in one
go — long-press to multi-select in Android's picker.
**Pass:** two rows appear in the queue with the names the picker showed, and both land in the directory the
breadcrumb names. The pane's listing shows them after **↻**.
**Failure means:** one row from a two-document pick is `PickAsync` losing the rest, and two rows with one
name is the per-file staging directory having gone — that is the overwrite `DocumentStaging` documents, and
it silently uploads the same bytes twice.
### 14.2 The name that arrives is the name that was picked · **the one most likely to be wrong**
Pick a document whose display name has a space and a non-Latin character in it, and one from a cloud
provider — Drive, or the Downloads shortcut — rather than local storage.
**Pass:** the file on the host is called what the picker called it. A cloud document uploads too, or fails
with the provider's own message in the status line rather than a crash.
**Failure means:** a mangled name is `SafeName` over-reaching. A name that reaches the host with a `/` in it
is `SafeName` under-reaching, and that one writes to a path nobody chose. A cloud document that hangs is
the copy being made on the interface thread — the whole reason `CopyInAsync` leaves it.
### 14.3 The queue is bounded, and the buttons stay reachable
Queue five or six files at once, on a small phone if there is one.
**Pass:** the queue scrolls inside its own region and **ADD FILES**, **DELETE** and **CLOSE** are all still
on screen. Every button is a thumb's size.
**Failure means:** buttons pushed off the bottom is the `MaxHeight` gone from the queue's scroller, and it
makes the screen unusable exactly when somebody has queued the most work.
### 14.4 A stopped upload resumes rather than starting again
Start a large upload, press **STOP** part way, then press **RESUME**.
**Pass:** it carries on from roughly where it stopped rather than from zero — the progress text is the thing
to read.
**Failure means:** restarting from zero means the staged copy was deleted at the stop, which is precisely
what `QueueStagedUploads` does not do and why it does not. A failure saying the file cannot be found is the
same bug, one step further along.
### 14.5 The copies do not accumulate · **the one nothing else would catch**
Note the app's storage in Android Settings → Apps → DodoSSH → Storage. Upload a large file, let it finish,
and look again. Then stop an upload part way, leave it stopped, force-stop the app and relaunch it.
**Pass:** storage returns to about what it was after the successful upload — the copy is deleted the moment
the transfer completes. After the stopped one, the cache is bigger while the app stays open (the copy is
being kept for RESUME) and back to its old size after the relaunch, which is `DocumentStaging.Sweep`.
**Failure means:** growth after a successful upload is `ReleaseStaged` not firing, and every file sent
leaves a second copy on the phone until Android reclaims the cache. Growth that survives a relaunch is the
sweep not running.
### 14.6 The notification is up while it transfers, and gone afterwards
Queue several files in each direction, put the phone to sleep with the screen off, and wait.
**Pass:** the foreground notification is up, the transfers finish while the screen is off, and the
notification goes away when the last one does — with no shell open. With a shell open it stays, because that
is what it was already for.
**Failure means:** an upload that stalls with the screen off is the count not reaching
`SessionForegroundService`, and Android has stopped the process mid-transfer. A notification left up
afterwards is `ActivityChanged` not being subscribed — the other end of the same wire.
### 14.7 SAVE FILE writes where you pointed it, and the file opens
Select a file on the host — something with a viewer, an image or a PDF — press **SAVE FILE**, and put it
somewhere reachable: Downloads, or a folder in Drive. When the transfer finishes, open it from the phone's
own Files app.
**Pass:** the status line says it was saved, the file is where the picker was pointed under the name shown
there, and it opens with the right contents. The queue row says DONE.
**Failure means:** a row that says DONE with nothing at the destination is `DeliverAsync` never running —
the delivery is registered per transfer id, and losing it makes the download look like a success while the
bytes sit in a cache nobody can reach. A file that is there but empty or truncated is the copy out, not the
transfer: check the `SetLength(0)` and that the write stream is being disposed before the status is written.
### 14.8 The button is dead until a file is chosen, and refuses a directory
With nothing selected, look at **SAVE FILE**. Then select a directory row.
**Pass:** disabled in both cases — it needs a connected remote and a selected *file*, which is the desktop's
own `CanDownload`.
**Failure means:** an enabled button over a directory reaches `QueueDeliveredDownload`'s refusal, which is
the right answer arriving too late: the save picker has already created an empty file, so the person is left
with a file they did not want and a message saying nothing happened.
### 14.9 A download that fails leaves the empty file it warned about
Point SAVE FILE at a destination for a large file, then break the transfer — turn off Wi-Fi and mobile data
while it runs.
**Pass:** the row goes to FAILED with the reason, the status line does not claim it was saved, and there is
an empty file at the destination. Reconnect, press **RETRY**, and the same destination fills in — the
delivery survives the failure because it is held against the transfer rather than the attempt.
**Failure means:** a retry that succeeds but leaves the destination empty is the delivery having been
dropped on the failure. An error saying the staged file is missing is the copy having been deleted at the
stop, which is what `QueueDeliveredDownload` documents it does not do.