Let a shared vault arrive, a bucket be found, and a vault be deleted

Three things a user reported, one of which was a real bug and one of which was
not the bug it looked like.

**A vault shared with somebody never reached their machine.** The grant was
correct at both ends: the sharing client verified the recipient's key against the
key log and wrapped every generation to it, the server stored it, and /me would
have returned it. Nothing asked. VaultSession.RefreshVaultsAsync — the method
whose own summary says it is "called after a share and on a periodic pass" — had
no caller anywhere in the application, so the vault list was whatever the last
browser sign-in cached. A restart did not help: an offline unlock reads that same
cache. The vault appeared only if the recipient happened to sign in through the
browser again, which is why this looked like sharing being broken rather than
like a list that was never re-read.

So every synchronisation pass now re-reads it, before it syncs. SyncOnceAsync
takes the whole server rather than its sync half for that reason, and the order
matters: a vault admitted by the refresh is one that same pass then pulls, where
the other order would show a newly shared vault as an empty one until the minute
after. The shell is told only when the set actually changed — it rebuilds the tab
strip's vault menu from the session's list, and doing that on every quiet pass
would rebuild a menu once a minute for nothing.

The test needed the fake server to be able to do something no test here had
needed before: hand this account a vault it did not make. ShareVaultWithMe wraps
a real key to the encryption key this account enrolled, so the keyring opens it
exactly as it opens a real colleague's — a helper that filled the field with
bytes would let a vault appear in the list and never prove it could be read.

**Adding an S3 bucket on the desktop works, and could not be found.** The report
was that it is not possible; driving the real XAML headlessly says otherwise —
Keychain, + BUCKET, and the editor saves. What is true is that S3 is where
somebody goes looking, and from there SELECT BUCKET opened a combo box with
nothing in it and no sentence anywhere saying that a bucket is a keychain item.
From where the user was standing that is indistinguishable from an application
with no way to add one.

The empty state now says what a bucket is and offers a button that lands on the
keychain with the editor already open — navigating to the screen and leaving
+ BUCKET to be found among five buttons would be most of the same problem. The
phone gets the sentence and no button: its keychain screen reads and deletes and
edits nothing, so there is no editor to send anybody to, and naming the machine
that has one beats an empty control that reads as a screen still loading.

The keychain screen's layout test grew the two categories it never covered.
Tags and buckets arrived after it was written, and the header strip it measures
is one that has overflowed twice before.

**A vault can now be deleted.** DELETE /api/v1/vaults/{id}, gated on Admin —
the line the rename already drew, for a stronger version of its reason, since
this takes the vault from everybody in it at once. The row is soft-deleted and
every grant to it withdrawn in one write; VaultAccessService filters on the stamp
at both ends, so from that moment the vault is absent from every member's /me and
every call naming it answers 404. Their clients notice on the pass described
above.

The team behind it is archived when it owned nothing else, which is the mirror of
renaming it: a vault made from the vaults screen gets a team named after it that
nobody was ever shown, and leaving that behind would leave a membership list no
screen has a row for. That is a second call rather than one transaction —
archiving is TeamService's, it refuses while a team owns vaults, and it can only
tell that this one no longer does once the deletion is committed. A crash between
the two leaves an empty team: invisible, archivable afterwards, harmless, and a
better failure than a vault that could not be deleted because tidying up after it
did not work.

Two refusals worth stating. The personal vault cannot be deleted at either end:
it is created by enrollment, everything filed nowhere else lives in it, and no
call would make another. And the items are kept — ciphertext behind a vault
nothing will resolve, so deleting them buys no confidentiality while destroying
what an operator undoing a mistake would need.

The client drops the key from the keyring and the row from the cache rather than
waiting for a refresh, so the list is right immediately; the items stay, as they
stay for a vault whose grant was withdrawn, because a copy is on every other
member's machine too and removing these rows would be the client pretending to a
reach it does not have. The confirmation says that out loud before it is
answered. It is the one sentence this screen must not leave implied: deletion is
no more retroactive than revocation is. See ADR 0001.

Desktop only, deliberately. The Android vaults screen offers no rename and no
hand-over either, so adding delete alone there would be the one destructive vault
operation on a screen with no other.

Three places asserted that a vault can never be deleted — TeamService's refusal
message, the TeamNotEmpty problem code, and ADR 0009 — and each now names the
route instead.

src/DodoSSH.Api/Features/Teams/TeamLog.cs

@@ -60,6 +60,19 @@ internal static partial class TeamLog
         Message = "Renamed vault {VaultId} of team {TeamId}.")]
     internal static partial void VaultRenamed(ILogger logger, Guid vaultId, Guid? teamId);
 
+    /// <remarks>
+    /// A warning rather than information, and the count is why: this is the one operation that takes a
+    /// vault away from everybody at once, and how many keys were withdrawn with it is the number an
+    /// operator asked about it will want. No name, for the reason the rename gives.
+    /// </remarks>
+    [LoggerMessage(
+        EventId = 2116,
+        Level = LogLevel.Warning,
+        Message = "Deleted vault {VaultId} of team {TeamId}, by {ActorId}; withdrew {GrantCount} key "
+            + "grant(s). Its items are kept and unreadable; copies already synced elsewhere are unaffected.")]
+    internal static partial void VaultDeleted(
+        ILogger logger, Guid vaultId, Guid teamId, Guid actorId, int grantCount);
+
     [LoggerMessage(
         EventId = 2106,
         Level = LogLevel.Information,

src/DodoSSH.Api/Features/Teams/TeamService.cs

@@ -218,10 +218,15 @@ internal sealed class TeamService(
     /// <b>The vault check is the whole of this operation's safety and it refuses rather than
     /// cascades.</b> Archiving a team hides it from every member's list at once, and a team vault
     /// resolves through membership — so archiving one that still owned vaults would take those vaults
-    /// away from people who hold keys to them, silently, including the caller. Nothing in this product
-    /// deletes a vault, so there is no sequence of calls that turns this refusal into a success today.
-    /// That is stated plainly rather than worked around, for the reason the SFTP layer refuses a
-    /// recursive delete: a refusal is visible and a quiet removal is not.
+    /// away from people who hold keys to them, silently, including the caller. The way out is to delete
+    /// those vaults first — <c>VaultGrantService.DeleteVaultAsync</c>, which asks its own question and
+    /// withdraws every key — and then archive what is left. A refusal that names a route is worth more
+    /// than a cascade, for the reason the SFTP layer refuses a recursive delete: a refusal is visible and
+    /// a quiet removal is not.
+    /// </para>
+    /// <para>
+    /// Deleting the <em>last</em> vault of a team made to carry it archives that team on the way past, so
+    /// the ordinary case never reaches this refusal at all. See <c>DeleteVaultEndpoint</c>.
     /// </para>
     /// <para>
     /// Memberships are archived with the team, in one transaction, because a live membership pointing
@@ -246,7 +251,7 @@ internal sealed class TeamService(
             throw new TeamNotEmptyException(
                 string.Create(
                     CultureInfo.InvariantCulture,
-                    $"This team still owns {vaultCount} vault(s), and archiving it would take them away from everybody holding a key — including you. There is no way to delete a vault in this product yet, so a team with vaults cannot be archived."));
+                    $"This team still owns {vaultCount} vault(s), and archiving it would take them away from everybody holding a key — including you. Delete those vaults first, which asks about each one and withdraws every key to it."));
         }
 
         var now = clock.GetUtcNow();

src/DodoSSH.Api/Features/Teams/VaultGrantEndpoints.cs

@@ -122,6 +122,96 @@ internal sealed class RenameVaultEndpoint(
     }
 }
 
+/// <summary>Deletes a vault.</summary>
+/// <remarks>
+/// <para>
+/// Admin, like the rename above and for a stronger version of the same reason: this takes the vault away
+/// from everybody in it at once, so it belongs to whoever administers the membership list rather than to
+/// anybody who may write to its contents.
+/// </para>
+/// <para>
+/// Authenticated rather than Enrolled, again like the rename. Deleting a vault touches no key material —
+/// it withdraws grants rather than wrapping anything — so requiring a published identity key would refuse
+/// an admin for a reason that has nothing to do with what they are asking.
+/// </para>
+/// <para>
+/// <b>The team behind it is archived in a second call, and the two are deliberately not one
+/// transaction.</b> Archiving is <c>TeamService</c>'s, it refuses while a team owns vaults, and it can
+/// only tell that this one no longer does once the deletion is committed. A crash between the two leaves
+/// an empty team — invisible on every screen, archivable afterwards by the endpoint that exists for it,
+/// and harmless — which is a better failure than a vault that could not be deleted because tidying up its
+/// team failed.
+/// </para>
+/// </remarks>
+internal sealed class DeleteVaultEndpoint(
+    ICurrentUserContext currentUser,
+    IVaultAccessService vaultAccess,
+    VaultGrantService grants,
+    TeamService teams)
+    : EndpointWithoutRequest<Results<NoContent, NotFound, ProblemHttpResult>>
+{
+    /// <inheritdoc />
+    public override void Configure()
+    {
+        Delete("/api/v1/vaults/{vaultId:guid}");
+
+        Policies(Auth.AuthenticatedPolicy);
+
+        Description(b => b
+            .WithName("DeleteVault")
+            .WithSummary("Deletes a vault and withdraws every key to it.")
+            .WithTags("Vaults"));
+    }
+
+    /// <inheritdoc />
+    public override async Task<Results<NoContent, NotFound, ProblemHttpResult>> ExecuteAsync(
+        CancellationToken ct)
+    {
+        var user = await currentUser.GetOrProvisionAsync(ct).ConfigureAwait(false);
+        var access = await vaultAccess
+            .ResolveAsync(user.Id, Route<Guid>("vaultId"), ct)
+            .ConfigureAwait(false);
+
+        if (!access.Granted || !access.Permissions.HasFlag(PermissionFlags.Read))
+        {
+            // Indistinguishable from a vault that never existed, and from one somebody else deleted a
+            // moment ago — which is what makes a repeat of this call safe to send.
+            return TypedResults.NotFound();
+        }
+
+        if (!access.Permissions.HasFlag(PermissionFlags.Admin))
+        {
+            return Problems.Coded(
+                StatusCodes.Status403Forbidden,
+                ProblemCodes.Forbidden,
+                "Only an admin or the owner of the team that owns this vault can delete it.");
+        }
+
+        try
+        {
+            var orphaned = await grants.DeleteVaultAsync(user, access.Vault!, ct).ConfigureAwait(false);
+
+            if (orphaned is not null)
+            {
+                await teams.ArchiveAsync(user, orphaned, ct).ConfigureAwait(false);
+            }
+
+            return TypedResults.NoContent();
+        }
+        catch (VaultGrantInvalidException exception)
+        {
+            return Problems.Coded(
+                StatusCodes.Status400BadRequest, ProblemCodes.InvalidVaultGrant, exception.Message);
+        }
+        catch (TeamNotEmptyException)
+        {
+            // The team gained a vault between the delete and the archive. The deletion stands and is what
+            // was asked for; the team staying is not a failure the caller can or should act on.
+            return TypedResults.NoContent();
+        }
+    }
+}
+
 /// <summary>Wraps this vault's key to another member.</summary>
 /// <remarks>
 /// The one call in this API whose body the server can neither produce nor check. It stores a sealed

src/DodoSSH.Api/Features/Teams/VaultGrantService.cs

@@ -216,6 +216,86 @@ internal sealed class VaultGrantService(
             RekeyRequired: vault.RekeyRequired);
     }
 
+    /// <summary>
+    /// Deletes a vault, and the team behind it where that team exists to carry this vault alone.
+    /// </summary>
+    /// <returns>
+    /// The team that has been left owning nothing by this, or null where there is none. The caller
+    /// archives it; see <see cref="DeleteVaultEndpoint"/> for why that is a second step.
+    /// </returns>
+    /// <remarks>
+    /// <para>
+    /// <b>Soft, like every other removal here.</b> The row is stamped rather than dropped, so an operator
+    /// can see that a vault existed and what became of it — and so the foreign keys from its items and its
+    /// grants stay valid. <c>VaultAccessService</c> filters on the stamp at both ends, so from the moment
+    /// this commits the vault is absent from every member's <c>/me</c> and every call naming it answers 404.
+    /// </para>
+    /// <para>
+    /// <b>The items are deliberately left alone.</b> They are ciphertext behind a vault nothing will now
+    /// resolve, so deleting them buys no confidentiality — and it would destroy the one thing an operator
+    /// restoring a vault deleted by mistake would need. What this is not is a promise about other people's
+    /// machines: a member who synced yesterday still holds their copy, exactly as ADR 0001 says about
+    /// revocation, and the deletion message says so.
+    /// </para>
+    /// <para>
+    /// <b>Every grant is withdrawn in the same write.</b> Nothing reads them once the vault is gone, but a
+    /// live grant on a deleted vault is a row that says somebody holds a key to something that no longer
+    /// exists — and the grant list is the thing an operator reads to answer "who could open this".
+    /// </para>
+    /// </remarks>
+    internal async Task<Team?> DeleteVaultAsync(
+        UserAccount actor,
+        Vault vault,
+        CancellationToken cancellationToken)
+    {
+        ArgumentNullException.ThrowIfNull(vault);
+
+        // A personal vault is where everything filed nowhere else lives, and it is created by enrollment
+        // rather than by anybody choosing to make it. Deleting one would leave an enrolled account with a
+        // key, no vault, and no way to make the vault it is supposed to have.
+        if (vault.OwnerKind != VaultOwnerKind.Team || vault.TeamId is not { } teamId)
+        {
+            throw new VaultGrantInvalidException(
+                "A personal vault cannot be deleted. It is where everything filed nowhere else lives, and "
+                + "the account has no way to make another. Move what you want to keep into a shared vault "
+                + "and delete that instead.");
+        }
+
+        var now = clock.GetUtcNow();
+
+        var grants = await database.VaultKeyGrants
+            .Where(g => g.VaultId == vault.Id && g.RevokedAtUtc == null)
+            .ToListAsync(cancellationToken)
+            .ConfigureAwait(false);
+
+        foreach (var grant in grants)
+        {
+            grant.State = GrantState.Revoked;
+            grant.RevokedAtUtc = now;
+        }
+
+        vault.DeletedAtUtc = now;
+        vault.UpdatedAtUtc = now;
+
+        await database.SaveChangesAsync(cancellationToken).ConfigureAwait(false);
+
+        TeamLog.VaultDeleted(logger, vault.Id, teamId, actor.Id, grants.Count);
+
+        // Read after the save, so "owns nothing else" is asked of a database that already knows this one
+        // is gone. The team is returned rather than archived here for the reason the rename gives about
+        // renaming it: a team carrying several vaults has a name and a membership list somebody chose, and
+        // one made to carry this vault alone is a thing its creator was never shown.
+        var alone = !await database.Vaults
+            .AnyAsync(other => other.TeamId == teamId && other.DeletedAtUtc == null, cancellationToken)
+            .ConfigureAwait(false);
+
+        return alone
+            ? await database.Teams
+                .SingleOrDefaultAsync(t => t.Id == teamId && t.DeletedAtUtc == null, cancellationToken)
+                .ConfigureAwait(false)
+            : null;
+    }
+
     /// <summary>Lists who can open a vault.</summary>
     /// <remarks>
     /// One row per holder, not one per grant. A rotated vault holds several grants per member — one per

src/DodoSSH.Api/Setup/EndpointRegistration.cs

@@ -57,6 +57,7 @@ internal static class EndpointRegistration
             typeof(RevokeTeamInvitationEndpoint),
             typeof(CreateTeamVaultEndpoint),
             typeof(RenameVaultEndpoint),
+            typeof(DeleteVaultEndpoint),
             typeof(ListVaultGrantsEndpoint),
             typeof(IssueVaultGrantEndpoint),
             typeof(RevokeVaultGrantEndpoint),

src/DodoSSH.Client.Android/Views/FilesScreen.axaml

@@ -85,7 +85,7 @@
           </ComboBox.ItemTemplate>
         </ComboBox>
 
-        <ComboBox HorizontalAlignment="Stretch" MinHeight="44" IsVisible="{Binding ShowsBucketPicker}"
+        <ComboBox HorizontalAlignment="Stretch" MinHeight="44" IsVisible="{Binding ShowsBucketChoice}"
                   ItemsSource="{Binding Buckets}" SelectedItem="{Binding SelectedBucket}"
                   PlaceholderText="choose a bucket">
           <ComboBox.ItemTemplate>
@@ -98,6 +98,16 @@
           </ComboBox.ItemTemplate>
         </ComboBox>
 
+        <!--
+          A picker with nothing in it says nothing, and here it cannot say much more: this head's keychain
+          screen reads and deletes and edits nothing, so there is no bucket editor to send anybody to. The
+          sentence is therefore the whole of what this can offer, and it names the machine that can — which
+          is better than an empty control that reads as a screen still loading. The desktop's counterpart
+          has a button, because over there the editor is two clicks away.
+        -->
+        <TextBlock Classes="detail" TextWrapping="Wrap" IsVisible="{Binding ShowsNoBuckets}"
+                   Text="No buckets yet. A bucket is a keychain item — its name, keys and endpoint — and this phone can read the keychain but not add to it. Add one on the desktop and it syncs here." />
+
         <!--
           Its own box, deliberately not the hosts screen's. This is a separate authentication, so a password
           typed to open a terminal has not been offered here — and quietly reusing it would make a one-time
@@ -106,8 +116,13 @@
         <TextBox Classes="field secret" IsVisible="{Binding SelectedHostAsksForAPassword}"
                  Text="{Binding TypedPassword}" PlaceholderText="password" />
 
+        <!--
+          Gone rather than disabled when there is no bucket to open, because a disabled OPEN under an empty
+          picker is the same dead end with a grey button on it. The sentence above is what the card carries
+          in that state.
+        -->
         <Button Classes="primary" Content="{Binding ConnectLabel}" Command="{Binding ConnectCommand}"
-                IsEnabled="{Binding !IsBusy}" />
+                IsVisible="{Binding !ShowsNoBuckets}" IsEnabled="{Binding !IsBusy}" />
       </StackPanel>
     </Border>
 

src/DodoSSH.Client.Api/DodoSshApiClient.cs

@@ -202,6 +202,24 @@ public interface IVaultGrantApi
         UpdateVaultRequest request,
         CancellationToken cancellationToken);
 
+    /// <summary>
+    /// Deletes a vault and withdraws every key to it.
+    /// </summary>
+    /// <returns>Whether there was a vault to delete.</returns>
+    /// <remarks>
+    /// <para>
+    /// False rather than an exception for a vault that is already gone, exactly as
+    /// <see cref="RevokeVaultGrantAsync"/> answers about a grant: a caller driving towards "this vault is
+    /// no longer there" has arrived, and two admins deleting the same vault must not leave the slower one
+    /// looking at an error about something that happened.
+    /// </para>
+    /// <para>
+    /// What it does not do is reach anybody's machine. A member who synced before this holds their copy
+    /// afterwards — see ADR 0001 on revocation — and the interface that offers this has to say so.
+    /// </para>
+    /// </remarks>
+    Task<bool> DeleteVaultAsync(Guid vaultId, CancellationToken cancellationToken);
+
     /// <summary>Lists who holds a key to this vault.</summary>
     Task<VaultGrantsResponse> ListVaultGrantsAsync(Guid vaultId, CancellationToken cancellationToken);
 
@@ -599,6 +617,12 @@ public sealed class DodoSshApiClient(HttpClient http, IAccessTokenProvider token
             DodoSshJsonContext.Default.VaultSummary,
             cancellationToken);
 
+    /// <inheritdoc />
+    public Task<bool> DeleteVaultAsync(Guid vaultId, CancellationToken cancellationToken) =>
+        DeleteAsync(
+            string.Create(CultureInfo.InvariantCulture, $"/api/v1/vaults/{vaultId}"),
+            cancellationToken);
+
     /// <inheritdoc />
     public Task<VaultGrantsResponse> ListVaultGrantsAsync(
         Guid vaultId,

src/DodoSSH.Client.App/Views/TransfersScreen.axaml

@@ -407,7 +407,15 @@
             <TextBlock Classes="heading" FontSize="16" TextAlignment="Center"
                        Text="Connect to a host" IsVisible="{Binding ShowsHostPicker}" />
             <TextBlock Classes="heading" FontSize="16" TextAlignment="Center"
-                       Text="Open a bucket" IsVisible="{Binding ShowsBucketPicker}" />
+                       Text="Open a bucket" IsVisible="{Binding ShowsBucketChoice}" />
+            <!--
+              A different heading rather than the same one over an empty picker, which is what this screen
+              used to do: SELECT BUCKET opened a combo box with nothing in it and no sentence anywhere saying
+              where a bucket comes from. From here that is indistinguishable from an application that cannot
+              add one — and this is the screen somebody looking for S3 arrives at.
+            -->
+            <TextBlock Classes="heading" FontSize="16" TextAlignment="Center"
+                       Text="No buckets yet" IsVisible="{Binding ShowsNoBuckets}" />
 
             <!--
               The sentence, and the button under it. Both go away once the picker is up — by then they have
@@ -425,14 +433,26 @@
                          IsVisible="{Binding ShowsHostPicker}"
                          Text="Browsing a host's files opens its own SFTP connection, so the host records a second login — it is not the channel a terminal uses." />
               <TextBlock Classes="hint" FontSize="12" TextAlignment="Center"
-                         IsVisible="{Binding ShowsBucketPicker}"
+                         IsVisible="{Binding ShowsBucketChoice}"
                          Text="A bucket has nothing to connect to: the keys and the endpoint are tested by the first listing, not by opening a session." />
+              <!--
+                Where a bucket comes from, said on the screen that needs one. It names the keychain rather
+                than only offering a button, because the button is a shortcut and the sentence is what makes
+                the bucket findable again next week without it.
+              -->
+              <TextBlock Classes="hint" FontSize="12" TextAlignment="Center"
+                         IsVisible="{Binding ShowsNoBuckets}"
+                         Text="A bucket is an item in your keychain — the bucket name, its keys and its endpoint, encrypted like everything else there. Add one and it appears here." />
               <Button Classes="accent" HorizontalAlignment="Center" Content="SELECT HOST"
                       IsVisible="{Binding ShowsHostPicker}"
                       Command="{Binding BeginChoosingRemoteCommand}" />
               <Button Classes="accent" HorizontalAlignment="Center" Content="SELECT BUCKET"
-                      IsVisible="{Binding ShowsBucketPicker}"
+                      IsVisible="{Binding ShowsBucketChoice}"
                       Command="{Binding BeginChoosingRemoteCommand}" />
+              <Button Classes="accent" HorizontalAlignment="Center" Content="+ BUCKET"
+                      IsVisible="{Binding ShowsNoBuckets}"
+                      Command="{Binding AddBucketCommand}"
+                      ToolTip.Tip="Goes to the keychain with the bucket editor open." />
             </StackPanel>
 
             <!-- ==== The picker, once it has been asked for ==== -->
@@ -454,9 +474,14 @@
                 </ComboBox.ItemTemplate>
               </ComboBox>
 
+              <!--
+                Gated on there being one, not merely on this being the S3 screen. An empty picker is the
+                thing that made this screen look like it could not add a bucket, and the invitation above
+                now sends somebody to the keychain instead of opening this.
+              -->
               <ComboBox HorizontalAlignment="Stretch" ItemsSource="{Binding Buckets}"
                         SelectedItem="{Binding SelectedBucket}"
-                        IsVisible="{Binding ShowsBucketPicker}"
+                        IsVisible="{Binding ShowsBucketChoice}"
                         PlaceholderText="choose a bucket">
                 <ComboBox.ItemTemplate>
                   <DataTemplate x:DataType="vm:ObjectStoreRowViewModel">

src/DodoSSH.Client.App/Views/VaultsScreen.axaml

@@ -127,6 +127,11 @@
             The vault's own operations. RENAME is an admin's; handing it on is the owner's alone, and
             that is the line the server draws as well — an admin the owner promoted must not be able to
             take the vault from them.
+
+            DELETE is an admin's too, and it is last and red because it is the only one of the three that
+            cannot be undone. It is absent on the personal vault rather than disabled: there is no version
+            of this window in which that vault can go, so a greyed button would be an offer that never
+            becomes real. Its question is the card below, as every destructive answer on this screen is.
           -->
           <StackPanel Grid.Column="1" Orientation="Horizontal" Spacing="6"
                       IsVisible="{Binding ShowsVaultActions}">
@@ -136,6 +141,9 @@
             <Button Classes="ghost" Content="HAND OVER" Command="{Binding HandOverCommand}"
                     IsEnabled="{Binding !IsBusy}" IsVisible="{Binding OwnsSelected}"
                     ToolTip.Tip="Hands this vault to the selected member. They become its owner and you become an admin; only the new owner can hand it on again." />
+            <Button Classes="danger" Content="DELETE" Command="{Binding DeleteVaultCommand}"
+                    IsEnabled="{Binding !IsBusy}" IsVisible="{Binding CanDeleteSelected}"
+                    ToolTip.Tip="Deletes this vault and withdraws everybody's key to it. It cannot reach a machine that has already synced it." />
           </StackPanel>
         </Grid>
       </Border>

src/DodoSSH.Client.Session/VaultSharing.cs

@@ -198,6 +198,50 @@ public sealed partial class VaultSession
         return stored;
     }
 
+    /// <summary>
+    /// Deletes a vault, here and on the server.
+    /// </summary>
+    /// <param name="api">The vault calls.</param>
+    /// <param name="vaultId">The vault to delete.</param>
+    /// <param name="cancellationToken">Cancellation token.</param>
+    /// <returns>Whether there was a vault to delete.</returns>
+    /// <remarks>
+    /// <para>
+    /// The server first, and this machine only if it agreed. The other order would take a vault off this
+    /// screen and leave it on everybody else's, which is the one outcome worse than the deletion failing:
+    /// the person who pressed it is then the only one who believes it is gone.
+    /// </para>
+    /// <para>
+    /// <b>The key is dropped from the keyring, and the items are not.</b> Dropping the key is what makes
+    /// this machine unable to read what is left, which is the honest end state — the ciphertext is still
+    /// in this cache, as it is in every other member's, and pretending otherwise by deleting the rows
+    /// would be claiming a reach this product does not have. See ADR 0001. The rows go with the next
+    /// wipe of the cache; nothing reads them meanwhile, because every list is built from the vaults the
+    /// keyring can open.
+    /// </para>
+    /// </remarks>
+    public async Task<bool> DeleteVaultAsync(
+        IVaultGrantApi api,
+        Guid vaultId,
+        CancellationToken cancellationToken)
+    {
+        ObjectDisposedException.ThrowIf(disposed, this);
+        ArgumentNullException.ThrowIfNull(api);
+
+        if (!await api.DeleteVaultAsync(vaultId, cancellationToken).ConfigureAwait(false))
+        {
+            return false;
+        }
+
+        keyring.Forget(vaultId);
+
+        await Vault.RemoveAsync(vaultId, cancellationToken).ConfigureAwait(false);
+
+        Vaults = await Vault.ListAsync(cancellationToken).ConfigureAwait(false);
+
+        return true;
+    }
+
     /// <summary>
     /// Wraps a vault's key to another member, after verifying their published key.
     /// </summary>

src/DodoSSH.Client.Shell/ViewModels/MainWindowViewModel.cs

@@ -377,7 +377,10 @@ internal sealed partial class MainWindowViewModel : ObservableObject, IAsyncDisp
         // to register for a reason no message could explain.
         this.deviceName = string.IsNullOrWhiteSpace(deviceName) ? Environment.MachineName : deviceName;
 
-        transfers = new TransfersViewModel(sftpSessions, clock);
+        // The fourth argument is the route out of the S3 screen's empty state. A bucket is made on the
+        // keychain screen, which is a different tab and two clicks away from somebody who has gone to S3
+        // to add one — so the screen that needs a bucket is given a way to reach the screen that makes one.
+        transfers = new TransfersViewModel(sftpSessions, clock, addBucket: ShowNewBucket);
 
         // Built once, like the workspace it writes for, and given a vault only while one is open. It has to
         // outlive every lock for the same reason the workspace does: a shell opened before a lock is still
@@ -1300,6 +1303,27 @@ internal sealed partial class MainWindowViewModel : ObservableObject, IAsyncDisp
         vaults.NewVaultCommand.Execute(null);
     }
 
+    /// <summary>
+    /// Goes to the keychain with the bucket editor open.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// The same shape as <see cref="ShowNewVault"/> and for the same reason: the thing being made lives on
+    /// one screen, and the moment somebody wants it happens on another. Here the two are a whole tab apart —
+    /// S3 is where a bucket is used and the keychain is where its keys are kept — which is what made the S3
+    /// screen's empty picker read as an application that could not add one at all.
+    /// </para>
+    /// <para>
+    /// Silent when the vault is locked, which is a state the button behind this is not reachable in: the
+    /// screen it sits on is inside the unlocked half of the window.
+    /// </para>
+    /// </remarks>
+    private void ShowNewBucket()
+    {
+        ShowScreen(ShellScreen.Keychain);
+        Vault?.NewObjectStoreCommand.Execute(null);
+    }
+
     // ---- The phone's connect menu ----
 
     /// <summary>
@@ -2599,6 +2623,8 @@ internal sealed partial class MainWindowViewModel : ObservableObject, IAsyncDisp
             // "connecting…" for ever and leave the session it eventually opened with nothing in the window
             // naming it, and so no way to close it. The subscription dies with the vault once the attempt
             // resolves: the vault holds the handler, not the other way round.
+            oldValue.VaultsChanged -= OnVaultsAdmitted;
+
             if (attempts.Count == 0)
             {
                 oldValue.ConnectionStarting -= OnVaultConnectionStarting;
@@ -2614,6 +2640,12 @@ internal sealed partial class MainWindowViewModel : ObservableObject, IAsyncDisp
             newValue.SessionOpened += OnVaultSessionOpened;
             newValue.PropertyChanged += OnVaultPropertyChanged;
 
+            // A vault somebody shared arrives on a synchronisation pass rather than through anything the
+            // user pressed here — so the menu that lists them is rebuilt from the event rather than at the
+            // end of a command, which is the one place a newly admitted vault has no command to be at the
+            // end of.
+            newValue.VaultsChanged += OnVaultsAdmitted;
+
             // The host list is rebuilt from scratch on every synchronisation pass, and a rebuilt row starts
             // disconnected — so without this the status dots go out once a minute underneath terminals that
             // are still open. The rows belong to the vault and the connection state belongs to the shell,
@@ -2696,6 +2728,16 @@ internal sealed partial class MainWindowViewModel : ObservableObject, IAsyncDisp
         AdoptTab(tab);
     }
 
+    /// <summary>
+    /// Redraws the vault menu after a synchronisation pass found a vault this account had not seen.
+    /// </summary>
+    /// <remarks>
+    /// The vault itself is already on every screen by the time this runs — the pass that admitted it
+    /// reloaded the lists — so this is only the one thing the vault view model has no way to reach: the
+    /// switches in the tab strip, which are built here from the session's vault list.
+    /// </remarks>
+    private void OnVaultsAdmitted(object? sender, EventArgs e) => RebuildVaultToggles();
+
     /// <remarks>
     /// The vault opens SSH sessions and this shell owns the strip they appear in, so this is the seam between
     /// them and nothing more.

src/DodoSSH.Client.Shell/ViewModels/TransfersViewModel.cs

@@ -313,13 +313,24 @@ internal sealed partial class TransfersViewModel : ObservableObject, IAsyncDispo
     /// removes the thread from the question rather than making the test guess it right.
     /// </para>
     /// </param>
+    /// <param name="addBucket">
+    /// Takes the user to where a bucket is made, or null in a head that has nowhere to take them.
+    /// <para>
+    /// A delegate rather than this screen making one itself, and it is the same seam the vaults screen uses
+    /// for the shell: a bucket is a keychain item, the editor that writes one belongs to the keychain screen,
+    /// and duplicating it here would be a second form writing the same secret. What this screen owns is the
+    /// knowledge that somebody standing on it with no buckets needs to be sent somewhere.
+    /// </para>
+    /// </param>
     internal TransfersViewModel(
         ISftpSessionFactory sftp,
         TimeProvider clock,
-        Action<Action>? post = null)
+        Action<Action>? post = null,
+        Action? addBucket = null)
     {
         this.sftp = sftp;
         this.post = post ?? (action => Dispatcher.UIThread.Post(action));
+        this.addBucket = addBucket;
 
         // The supplier answers with whatever session is current at the moment a transfer starts, which is
         // what lets a queue survive a disconnect and reconnect without every queued row failing.
@@ -332,6 +343,32 @@ internal sealed partial class TransfersViewModel : ObservableObject, IAsyncDispo
         RemoteEntries.CollectionChanged += (_, _) => OnPropertyChanged(nameof(HasRemoteEntries));
         LocalEntries.CollectionChanged += (_, _) => OnPropertyChanged(nameof(HasLocalEntries));
         Transfers.CollectionChanged += (_, _) => OnPropertyChanged(nameof(HasTransfers));
+
+        // The fourth, and it follows a list this screen does not own: the buckets are refilled from the
+        // vault on every synchronisation pass, so the invitation these flags decide between has to change
+        // with them rather than at the moment somebody pressed something. That is the path that matters —
+        // adding the first bucket happens on another screen, and this one has to notice when it comes back.
+        Buckets.CollectionChanged += (_, _) => RaiseBucketState();
+    }
+
+    /// <summary>Where a bucket is made, or null in a head with nowhere to send anybody.</summary>
+    private readonly Action? addBucket;
+
+    /// <summary>Goes to the keychain with the bucket editor open.</summary>
+    /// <remarks>
+    /// The dead end this replaces was a real one: SELECT BUCKET opened a picker with nothing in it, and
+    /// nothing anywhere on the screen said that a bucket is made on the keychain screen. Somebody who had
+    /// come to S3 to add one had arrived at the place it is used and been shown no route to the place it is
+    /// created.
+    /// </remarks>
+    [RelayCommand]
+    private void AddBucket()
+    {
+        // Closed first, so returning here from the keychain does not find a picker still open over a list
+        // that has since gained the bucket it was empty of.
+        IsChoosingRemote = false;
+
+        addBucket?.Invoke();
     }
 
     /// <summary>The hosts that can be connected to, which is the vault's list.</summary>
@@ -352,6 +389,28 @@ internal sealed partial class TransfersViewModel : ObservableObject, IAsyncDispo
     [ObservableProperty]
     private ObjectStoreRowViewModel? selectedBucket;
 
+    /// <summary>
+    /// Whether there is a bucket to open at all.
+    /// </summary>
+    /// <remarks>
+    /// The question this screen used to answer by showing an empty combo box. A bucket is made on the
+    /// keychain screen and nowhere else, and S3 is where somebody goes looking for it — so a picker with
+    /// nothing in it and no sentence beside it was, from where the user was standing, an application with no
+    /// way to add a bucket. See <see cref="AddBucket"/>.
+    /// </remarks>
+    internal bool HasBuckets => Buckets.Count > 0;
+
+    /// <summary>Whether this screen is on S3 with a bucket to offer.</summary>
+    /// <remarks>
+    /// The two flags are resolved here rather than combined in markup, because a binding cannot say "and" —
+    /// and because the pair is one question with two answers, which is easier to keep straight in one place
+    /// than in four <c>IsVisible</c> expressions that have to stay each other's opposites.
+    /// </remarks>
+    internal bool ShowsBucketChoice => ShowsBucketPicker && HasBuckets;
+
+    /// <summary>Whether this screen is on S3 with nothing to open yet.</summary>
+    internal bool ShowsNoBuckets => ShowsBucketPicker && !HasBuckets;
+
     /// <summary>
     /// Which sort of remote the right-hand pane is about to open.
     /// </summary>
@@ -1641,16 +1700,30 @@ internal sealed partial class TransfersViewModel : ObservableObject, IAsyncDispo
     partial void OnRemoteChanged(RemoteKind value)
     {
         OnPropertyChanged(nameof(ShowsHostPicker));
-        OnPropertyChanged(nameof(ShowsBucketPicker));
         OnPropertyChanged(nameof(ConnectLabel));
         OnPropertyChanged(nameof(SelectedHostAsksForAPassword));
 
+        RaiseBucketState();
+
         // Arriving at the other destination shows its invitation rather than a picker somebody left open on
         // this one — and the picker it would have left open is the wrong one, since the two kinds have
         // different lists behind them.
         IsChoosingRemote = false;
     }
 
+    /// <summary>Raises the three flags that answer "is there a bucket to open".</summary>
+    /// <remarks>
+    /// Together, because they are one fact read three ways and a caller that raised two of them would leave
+    /// an invitation and an empty state on screen at once.
+    /// </remarks>
+    private void RaiseBucketState()
+    {
+        OnPropertyChanged(nameof(ShowsBucketPicker));
+        OnPropertyChanged(nameof(HasBuckets));
+        OnPropertyChanged(nameof(ShowsBucketChoice));
+        OnPropertyChanged(nameof(ShowsNoBuckets));
+    }
+
     partial void OnIsConnectedChanged(bool value)
     {
         OnPropertyChanged(nameof(CanDownload));

src/DodoSSH.Client.Shell/ViewModels/VaultViewModel.cs

@@ -3845,7 +3845,7 @@ internal sealed partial class VaultViewModel(
                     return;
                 }
 
-                var report = await SyncOnceAsync(server.Sync, cancellationToken).ConfigureAwait(true);
+                var report = await SyncOnceAsync(server, cancellationToken).ConfigureAwait(true);
 
                 // Null means a background pass held the gate. Saying so beats reporting a sync that this
                 // press did not perform.
@@ -3935,7 +3935,7 @@ internal sealed partial class VaultViewModel(
                 return;
             }
 
-            var report = await SyncOnceAsync(server.Sync, cancellationToken).ConfigureAwait(true);
+            var report = await SyncOnceAsync(server, cancellationToken).ConfigureAwait(true);
 
             if (report is null)
             {
@@ -3971,12 +3971,19 @@ internal sealed partial class VaultViewModel(
     }
 
     /// <remarks>
+    /// <para>
     /// The gate is shared with the manual command, so a press and a tick can never overlap. Taken with a
     /// zero timeout rather than awaited: a pass that arrives while another is running has nothing to add by
     /// waiting for it, and queueing them would turn a slow server into a backlog of identical work.
+    /// </para>
+    /// <para>
+    /// Takes the whole server rather than its sync half, because the first thing a pass does is ask
+    /// <em>which vaults there are</em> — see <see cref="AdmitNewVaultsAsync"/>. A pass that only synced the
+    /// vaults it already knew could never discover one somebody had just shared.
+    /// </para>
     /// </remarks>
     private async Task<IReadOnlyList<VaultSyncReport>?> SyncOnceAsync(
-        ISyncApi api,
+        IVaultServer server,
         CancellationToken cancellationToken)
     {
         if (!await syncGate.WaitAsync(0, cancellationToken).ConfigureAwait(true))
@@ -3986,10 +3993,14 @@ internal sealed partial class VaultViewModel(
 
         try
         {
+            // Before the sync, so a vault admitted here is one of the vaults that pass then pulls. The
+            // other order would show a newly shared vault as an empty one until the minute after.
+            await AdmitNewVaultsAsync(server.Account, cancellationToken).ConfigureAwait(true);
+
             // Every vault this session can read, not only the one new items are filed into. A team's
             // vault that never synced would show its hosts exactly once — at the unlock that first
             // pulled it — and then quietly stop, which reads as the feature not working.
-            var report = await session.SyncAllAsync(api, cancellationToken).ConfigureAwait(true);
+            var report = await session.SyncAllAsync(server.Sync, cancellationToken).ConfigureAwait(true);
 
             // Not unconditionally false, which it was while a pass was one vault and a failure was an
             // exception. A failure is now a report — one unreachable team vault must not stop the others
@@ -4012,6 +4023,53 @@ internal sealed partial class VaultViewModel(
         }
     }
 
+    /// <summary>
+    /// Re-reads which vaults this account can reach, and opens any that have become readable.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// <b>This is the whole of how a shared vault arrives.</b> Sharing is two acts on two machines: the
+    /// person sharing wraps the vault key to the recipient, and the recipient's own client has to notice.
+    /// The recipient is handed nothing — there is no push channel — so without this the vault list stayed
+    /// exactly as it was cached at sign-in, and a vault shared with somebody appeared on their machine only
+    /// if they happened to sign in through the browser again. Everything else was already right, which is
+    /// why it looked like sharing was broken rather than like a list that was never re-read.
+    /// </para>
+    /// <para>
+    /// A failure is left to the caller, which treats it as the pass failing: the call is to the same server
+    /// the sync is about to use, so a refresh that cannot answer is not a state in which the sync would
+    /// have.
+    /// </para>
+    /// <para>
+    /// The shell is told only when the set actually changed. It rebuilds the tab strip's vault menu from
+    /// this list, and doing that on every quiet pass would rebuild a menu once a minute for nothing.
+    /// </para>
+    /// </remarks>
+    private async Task AdmitNewVaultsAsync(IAccountApi api, CancellationToken cancellationToken)
+    {
+        var before = session.Vaults.Count;
+
+        var admitted = await session.RefreshVaultsAsync(api, cancellationToken).ConfigureAwait(true);
+
+        if (admitted == 0 && session.Vaults.Count == before)
+        {
+            return;
+        }
+
+        VaultsChanged?.Invoke(this, EventArgs.Empty);
+    }
+
+    /// <summary>
+    /// Raised when a synchronisation pass found that the vaults this account can reach have changed.
+    /// </summary>
+    /// <remarks>
+    /// An event rather than a callback because the listener is the shell and the shell owns this object,
+    /// which is the same shape the connection events above use. What it is for is the tab strip's vault
+    /// menu: it is built from the session's vault list, so a vault admitted mid-session would otherwise be
+    /// on every screen and missing from the one control that can hide it.
+    /// </remarks>
+    internal event EventHandler? VaultsChanged;
+
     /// <remarks>
     /// <c>ConfigureAwait(true)</c> throughout, and that is load-bearing rather than habit: the loop is
     /// started from the UI thread, so every continuation returns to it and the observable collections

src/DodoSSH.Client.Shell/ViewModels/VaultsViewModel.cs

@@ -233,16 +233,38 @@ internal sealed record VaultInvitationRowViewModel(TeamInvitationSummary Invitat
 /// than read from the selection at confirm time — otherwise selecting a different row between arming
 /// and confirming would apply the answer to something else.
 /// </remarks>
+/// <param name="Kind">Which operation the answer applies to.</param>
 /// <param name="TeamId">The membership list the action is aimed at.</param>
-/// <param name="MemberId">The member it is aimed at.</param>
+/// <param name="MemberId">The member it is aimed at, or empty where it is not aimed at one.</param>
+/// <param name="VaultId">The vault it is aimed at.</param>
 /// <param name="Question">What is being asked.</param>
 /// <param name="Consequence">What will actually happen, stated honestly.</param>
 internal sealed record VaultActionRequest(
+    VaultActionKind Kind,
     Guid TeamId,
     Guid MemberId,
+    Guid VaultId,
     string Question,
     string Consequence);
 
+/// <summary>Which destructive operation a confirmation is standing in front of.</summary>
+/// <remarks>
+/// Carried on the request rather than inferred from which fields are set. There were two of these the
+/// moment deletion existed, and a confirm handler that guessed from a member id being empty would be one
+/// refactor away from carrying out the wrong one of the two.
+/// </remarks>
+internal enum VaultActionKind
+{
+    /// <summary>Not a legal value.</summary>
+    Unspecified = 0,
+
+    /// <summary>Hand the vault's membership list to somebody else.</summary>
+    HandOver = 1,
+
+    /// <summary>Delete the vault.</summary>
+    Delete = 2,
+}
+
 /// <summary>
 /// The vaults screen: which vaults there are, who is in each, and who holds a key to it.
 /// </summary>
@@ -414,6 +436,16 @@ internal sealed partial class VaultsViewModel(
     /// <summary>Whether the selected vault is the personal one.</summary>
     internal bool SelectedIsPersonal => SelectedVault?.IsPersonal == true;
 
+    /// <summary>
+    /// Whether the selected vault is one this account may delete.
+    /// </summary>
+    /// <remarks>
+    /// Narrower than <see cref="CanAdministerSelected"/> by exactly the personal vault, which cannot be
+    /// deleted by anybody: it is where everything filed nowhere else lives and nothing can make another.
+    /// The server refuses one too, so a button offered here would be a button that leads to a refusal.
+    /// </remarks>
+    internal bool CanDeleteSelected => SelectedVault is { IsPersonal: false, CanAdminister: true };
+
     /// <summary>Whether there is anything to show beside the vault list.</summary>
     internal bool HasSelection => SelectedVault is not null;
 
@@ -1230,13 +1262,70 @@ internal sealed partial class VaultsViewModel(
         }
 
         PendingAction = new VaultActionRequest(
+            VaultActionKind.HandOver,
             teamId,
             member.UserId,
+            vault.VaultId,
             $"Hand '{vault.Name}' to {member.Name}?",
             "They become its owner and you become an admin. You will not be able to take it back "
             + "yourself — only the new owner can hand it on. Your key to it is untouched.");
     }
 
+    /// <summary>
+    /// Arms the deletion confirmation for the selected vault.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// The consequence is spelled out at length rather than summarised, and every sentence in it is one
+    /// somebody could otherwise be surprised by afterwards. The last is the one this product must never
+    /// leave implied: deleting a vault does not reach the machines it has already synced to. That is the
+    /// same limit revocation has, for the same reason, and it is written down in ADR 0001.
+    /// </para>
+    /// <para>
+    /// The personal vault is refused here as well as by the server. A button that reached a refusal would
+    /// be teaching somebody to try things and read errors, and the reason is a fact this screen knows.
+    /// </para>
+    /// </remarks>
+    [RelayCommand]
+    private void DeleteVault()
+    {
+        if (SelectedVault is not { } vault)
+        {
+            return;
+        }
+
+        if (vault.IsPersonal)
+        {
+            Status = "Your personal vault cannot be deleted. It is where everything filed nowhere else "
+                + "lives, and there is no way to make another.";
+            return;
+        }
+
+        if (vault.TeamId is not { } teamId)
+        {
+            return;
+        }
+
+        var others = Members.Count(member => !member.IsSelf);
+
+        var shared = others > 0
+            ? string.Create(
+                CultureInfo.CurrentCulture,
+                $" {others} other member(s) lose it at the same moment, without being asked.")
+            : string.Empty;
+
+        PendingAction = new VaultActionRequest(
+            VaultActionKind.Delete,
+            teamId,
+            Guid.Empty,
+            vault.VaultId,
+            $"Delete '{vault.Name}'?",
+            $"Everything in it goes: its hosts, keys, passwords, snippets and buckets stop being readable "
+            + $"by anybody, including you, and nothing here can undo it.{shared} What it cannot do is "
+            + "reach a machine that has already synced this vault — a copy pulled yesterday is still "
+            + "there. Rotate the credentials that mattered.");
+    }
+
     /// <summary>Cancels an armed action.</summary>
     [RelayCommand]
     private void CancelAction() => PendingAction = null;
@@ -1258,22 +1347,71 @@ internal sealed partial class VaultsViewModel(
 
         PendingAction = null;
 
-        await RunAsync(async () =>
+        await RunAsync(() => request.Kind switch
         {
-            await server.Teams
-                .TransferTeamOwnershipAsync(
-                    request.TeamId,
-                    new TransferTeamOwnershipRequest(request.MemberId),
-                    cancellationToken)
-                .ConfigureAwait(true);
-
-            await ReloadAsync(cancellationToken).ConfigureAwait(true);
-
-            Status = "Handed over. You are an admin of this vault now, and only its new owner can hand "
-                + "it on again.";
+            VaultActionKind.Delete => DeleteAsync(server, request, cancellationToken),
+            _ => HandOverAsync(server, request, cancellationToken),
         }).ConfigureAwait(true);
     }
 
+    private async Task HandOverAsync(
+        IVaultServer server,
+        VaultActionRequest request,
+        CancellationToken cancellationToken)
+    {
+        await server.Teams
+            .TransferTeamOwnershipAsync(
+                request.TeamId,
+                new TransferTeamOwnershipRequest(request.MemberId),
+                cancellationToken)
+            .ConfigureAwait(true);
+
+        await ReloadAsync(cancellationToken).ConfigureAwait(true);
+
+        Status = "Handed over. You are an admin of this vault now, and only its new owner can hand "
+            + "it on again.";
+    }
+
+    /// <summary>
+    /// Deletes the vault the confirmation was armed for.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// The name is read before the call, because afterwards there is no row to read it from and the
+    /// sentence this ends with is about a vault that no longer exists.
+    /// </para>
+    /// <para>
+    /// The rest of the shell is told, as a rename tells it: the tab strip's vault menu, the file-this-into
+    /// picker and every host list are built from the session's vault list, and all of them are a vault out
+    /// of date the moment one goes.
+    /// </para>
+    /// </remarks>
+    private async Task DeleteAsync(
+        IVaultServer server,
+        VaultActionRequest request,
+        CancellationToken cancellationToken)
+    {
+        if (session() is not { } open)
+        {
+            Status = "Unlock your keychain first: deleting a vault gives up this machine's key to it.";
+            return;
+        }
+
+        var name = Vaults.FirstOrDefault(row => row.VaultId == request.VaultId)?.Name ?? "That vault";
+
+        var deleted = await open
+            .DeleteVaultAsync(server.Grants, request.VaultId, cancellationToken)
+            .ConfigureAwait(true);
+
+        await ReloadAsync(cancellationToken).ConfigureAwait(true);
+        await NotifyVaultsChangedAsync(cancellationToken).ConfigureAwait(true);
+
+        Status = deleted
+            ? $"Deleted '{name}'. Everybody's key to it is withdrawn. What anybody had already synced is "
+                + "still on their machine — rotate the credentials that mattered."
+            : $"'{name}' was already gone. Somebody else deleted it, or your access to it ended.";
+    }
+
     /// <summary>
     /// Removes somebody, revoking their grants and rotating the vaults they could read.
     /// </summary>
@@ -1629,6 +1767,7 @@ internal sealed partial class VaultsViewModel(
         OnPropertyChanged(nameof(OwnsSelected));
         OnPropertyChanged(nameof(SelectedIsShared));
         OnPropertyChanged(nameof(SelectedIsPersonal));
+        OnPropertyChanged(nameof(CanDeleteSelected));
         OnPropertyChanged(nameof(HasInvitations));
         OnPropertyChanged(nameof(SharedMembershipWarning));
         OnPropertyChanged(nameof(HasSharedMembershipWarning));

src/DodoSSH.Client.Storage/VaultStore.cs

@@ -146,6 +146,44 @@ public sealed class VaultStore(IDbContextFactory<ClientCacheContext> contexts, T
         await context.SaveChangesAsync(cancellationToken).ConfigureAwait(false);
     }
 
+    /// <summary>
+    /// Removes one vault's row and its wraps, leaving the rest alone.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// For the vault this machine has just deleted, which is gone here before the server's next
+    /// <c>/me</c> stops mentioning it — the mirror image of <see cref="UpsertAsync"/>, and needed for the
+    /// same reason: a screen that went on listing a vault the user has just deleted, until the next
+    /// refresh, would look as though the deletion had not worked.
+    /// </para>
+    /// <para>
+    /// The wraps go with it, as they do in <see cref="ReplaceAllAsync"/> and for the same reason: they are
+    /// keys to something that no longer exists. Its <em>items</em> stay, exactly as they stay for a vault
+    /// whose grant was withdrawn — the ciphertext is on every other member's machine too, and deleting
+    /// these rows would be the client pretending to a reach it does not have.
+    /// </para>
+    /// </remarks>
+    public async Task RemoveAsync(Guid vaultId, CancellationToken cancellationToken)
+    {
+        var context = contexts.CreateDbContext();
+        await using var scope = context.ConfigureAwait(false);
+
+        var row = await context.Set<CachedVaultRow>()
+            .SingleOrDefaultAsync(r => r.VaultId == vaultId, cancellationToken)
+            .ConfigureAwait(false);
+
+        if (row is null)
+        {
+            return;
+        }
+
+        context.Remove(row);
+
+        await RemoveWrapsAsync(context, vaultId, cancellationToken).ConfigureAwait(false);
+
+        await context.SaveChangesAsync(cancellationToken).ConfigureAwait(false);
+    }
+
     /// <summary>Reads the vaults this machine has been asked to leave off the screens.</summary>
     /// <remarks>
     /// The ids alone: a caller wanting the names already has <see cref="ListAsync"/>, and a list that

src/DodoSSH.Client.Sync/VaultKeyring.cs

@@ -228,6 +228,40 @@ public sealed class VaultKeyring : IDisposable
         }
     }
 
+    /// <summary>
+    /// Drops every key to one vault, zeroing them.
+    /// </summary>
+    /// <remarks>
+    /// <para>
+    /// For a vault that has been deleted, which is the one case where this session is not merely unable to
+    /// read a vault but has no vault to read. <see cref="MarkUnreadable"/> is the other shape and keeps the
+    /// keys deliberately — the items sealed under them are still there and still wanted — whereas here
+    /// there is nothing left for them to open and holding them would be keeping key material for something
+    /// that no longer exists.
+    /// </para>
+    /// <para>
+    /// It is also taken off the unopened list, which is what stops a deleted vault being reported for ever
+    /// as one this machine is waiting for a key to.
+    /// </para>
+    /// </remarks>
+    public void Forget(Guid vaultId)
+    {
+        ObjectDisposedException.ThrowIf(disposed, this);
+
+        generations.Remove(vaultId);
+        Unopened = [.. Unopened.Where(id => id != vaultId)];
+
+        if (!keys.Remove(vaultId, out var held))
+        {
+            return;
+        }
+
+        foreach (var key in held.Values)
+        {
+            CryptographicOperations.ZeroMemory(key);
+        }
+    }
+
     /// <summary>
     /// Borrows a vault's current key: the one new items are sealed under.
     /// </summary>

src/DodoSSH.Contracts/ProblemCodes.cs

@@ -112,10 +112,10 @@ public static class ProblemCodes
     /// A team cannot be archived while it still owns vaults.
     /// </summary>
     /// <remarks>
-    /// Its own code because the remedy is neither "fix what you typed" nor "pick another value": it is
-    /// to deal with the vaults first. Archiving anyway would hide vaults from every member including
-    /// the ones holding keys to them, and this product has no way to delete a vault, so the refusal is
-    /// the honest end of that road rather than a step on it.
+    /// Its own code because the remedy is neither "fix what you typed" nor "pick another value": it is to
+    /// deal with the vaults first, by deleting each of them — <c>DELETE /api/v1/vaults/{vaultId}</c> — which
+    /// asks about each one on the way. Archiving anyway would hide vaults from every member including the
+    /// ones holding keys to them, which is the same removal without anybody having been asked.
     /// </remarks>
     public const string TeamNotEmpty = "team-not-empty";