What you can rely on

CapabilityStatus
lore CLI, macOS / Linux / WindowsWorks today
CI / headless via API keyWorks today
Per-repo sharing with read / write / admin rolesWorks today
Partition read access enforced (non-members see nothing)Works today
Admin-gated destructive ops (obliterate, migrate locks)Works today
Disk cap enforced by the filesystem itselfWorks today
One lore version fleet-wide, upgraded for youWorks today
Push blocked for read-only membersNot yet - upstream lore TODO
VS Code / Unreal / Unity pluginsNot yet - upstream lore roadmap
Clone a repo by nameNot supported - clone by URL

Honest caveat: because push isn't yet role-gated upstream, treat read vs write as intent, not a hard push barrier today. Read access and the admin-only operations are enforced.

Your disk cap fails the way you'd want: writes stop with "no space left on device" at the limit, deletes always work, and freed space comes back immediately - never a read-only lockout.

Every workspace runs the same lore version - not a per-workspace choice. That's what lets us revoke a compromised token (lore has no other revocation mechanism) and move a workspace between hosts safely. When we upgrade the fleet, your workspace restarts briefly on its own - no data movement, nothing for you to do.