Systemiq structures enterprise context through a stack model made of client scope, systems, tools, and elements.

Records, metrics, summaries, and impacts are the derived outputs of that model. Operational memory is the persisted layer that keeps those outputs queryable over time.

Structural model

• Clients define the authenticated top-level boundary within which the model is interpreted and queried.
• Systems are major operational boundaries inside that client scope.
• Tools represent structured subsystems within a system.
• Elements are the atomic entities used to model state and relationships.
• Client-level outputs can interpret all or selected systems into higher-level insights and summaries.
• In causal connections and system-dynamics models, elements correspond to nodes and edges represent the connections between them.
• Tool-element mappings and impact records make subsystem structure and directional influence explicit across the stack.

What persists

• Records carry structured operational context and agent-produced outputs.
• Metrics carry numerical values for aggregation, trend analysis, and time-series querying.
• Summaries compress model state into reusable machine-consumable interpretations.
• Impacts capture how one layer influences another, such as element to tool, tool to system, or system to client-level interpretation.
• Together these outputs form the operational memory agents query later.

How agents use the model

• Use systems, tools, and elements when the agent first needs orientation inside the model.
• Use records when the task requires scoped operational context objects.
• Use metrics when the task requires numerical comparison, aggregation, or time-series analysis.
• Use indicators to address stable output families across the stack.
• Use mappings and impacts when the task depends on structure, dependencies, or directional influence.

How to scope queries

• Start broad only when discovering structure for the first time.
• Then narrow by `system_id`, `tool_id`, `element_id`, `indicator`, or `record_type`.
• Prefer task-shaped requests over broad raw-state retrieval.
• Use the reference page when you need exact query dimensions and field semantics.