Can a fleet of agents share one memory?

Yes. This is the fleet story: many agents, one shared memory spine, with governance attached to every write.

Without shared memory, a fleet is a collection of amnesiacs: each agent rediscovers what another learned an hour ago, and nothing joins up. With CueCrux, agents write facts, decisions and work state into one spine and retrieve from it under budgets, so the fleet behaves like one brain with many hands.

Sharing does not mean anonymity. Every write arrives through a passport-bound tool call and lands with a CROWN receipt, so shared memory stays attributable: you can always establish which agent contributed a fact, at what trust tier, and when. When something in the shared store is wrong, you trace it to its origin instead of debating it.

Continuity is built in. Sessions and checkpoints preserve working state; cross-session handoffs pass work between agents, or between an agent and a human, with the receipts attached; the live work board and ExecPlans keep plans, milestones and blockers where the whole fleet, and the humans running it, can see them.

Humans are first-class here. The operator console shows the same board, graph and sessions the agents act on, so oversight is not a separate reporting system that drifts out of date.

The hosted sync that puts one spine behind many machines arrives with Pro; the fleet-wide controls arrive with Governance.