# Glossary: Shared System Base, Bounded Extensions, Cases, Queues, And Shared Memory

The glossary defines the key nouns explicitly so agents and humans do not have to infer them from context or worldview writing.

## Summary

- Shared system base: The shared software base that multiple workflows and extensions rely on.
- Shared memory: Knowledge and durable context visible to teams and agents across workflows.
- Bounded extension: A focused capability slice that remains governed by the shared trust and operational model.
- Agent-visible collaboration: A workflow where agents can see the same durable records, context, and approvals as humans.

## Terms about the system itself.

These terms describe how Move Big Rocks is structured.

- Shared system base: The shared base of primitives and services that multiple workflows and extensions build on.
- Shared memory: Durable knowledge, policy, and context that remains available to teams and agents instead of disappearing into local files or private notes.
- Bounded extension: An extension that adds focused capability while still depending on shared workspaces, permissions, audit, and work records.
- Runtime services: The shared system services that support identity, routing, audit, visible records, and extension behavior.

## Terms about the operational work itself.

These terms describe how work moves through the system.

- Workspace: The primary operational and security boundary inside an instance.
- Queue: A visible bucket for work in motion with ownership and current operational state.
- Conversation: A live interaction record that can gather context, resolve directly, or lead into durable follow-through.
- Case: A durable operational record for ownership, follow-up, approvals, and auditable closure.
- Knowledge: Durable operational context such as runbooks, policies, templates, and shared guidance.
- Automation: System behavior that acts on explicit shared records instead of hidden side channels.

## Terms that matter during evaluation and adoption.

These terms describe how humans and agents are expected to use the system.

- Agent-visible collaboration: A workflow where agents can inspect and operate against the same records and knowledge people use.
- Human approval: A decision that should remain visible to a responsible person rather than being inferred or auto-expanded by an agent.
- Self-hosted production model: The long-term deployment posture in which the customer controls runtime, secrets, domains, and production responsibility.
- Sandbox: A disposable evaluation environment with the same broad model as production but a lower-trust and time-limited lifecycle.
- Authoritative surface: A public page or repository that is explicit about what it is canonical for, such as source, docs, deployment, or machine discovery.

## Related

- Agents: /agents
- Resources: /resources
- Philosophy: /philosophy