True iff this mount is ALSO in the app's own useMounts (the App lens);
false for a session-export-only mount visible only to the editor/agent + the
Session lens.
OptionalidOptional stable identifier (the spaceId, for spaces).
OptionalmodeAccess mode of the granted view: 'rw' (read-write) or 'ro' (read-only).
A live role downgrade re-announces the same mount with mode: 'ro'; apps
observing onMountsChange see the change and writes start failing EROFS.
Absent on the primary repo mount (treated as read-write).
OptionalnameHuman-readable label for the mount — the space's display name, or the repo
label for the primary working-tree mount (R3-69). Use this to show users and
agents what a mount is: the path (/mnt/{hash}) and id (the spaceId)
are opaque, and space names are not unique, so neither alone tells you which
filesystem you're looking at. Absent when the host can't resolve a name
(older host, or a name it never learned) — fall back to id/path.
Absolute path where the mount is reachable (e.g. /spaces/{id}).
OptionalrulesThe granted scopes of this mount (plan 12 §8.7 / §F): each {subtree, mode}
is a path prefix you hold and at what access, at the mount's backend-natural
paths. Use it to reason about per-path writability — which subtree is rw —
WITHOUT probing EROFS. A single whole-mount grant is [{ subtree: '/', mode }].
Absent on the primary repo mount and on an older host that doesn't report it.
Backend kind, e.g. 'firestore'.
A mount as seen through the first-party Session lens (PRINCIPALS_SPEC §9 B2): the session's mounts BEYOND this app's own (the editor/agent session's). This is a metadata view — no filesystem port — so it extends SandboxMount with only forwardedToApp.