Systemiq on GitHub

Reference

Concrete reference for MCP access, query scope, stack identifiers, records, and metrics.

MCP endpoint

  • Transport: HTTP MCP endpoint exposed by Systemiq.
  • Purpose: make operational context available to downstream agent clients.
  • Typical client configuration includes the MCP server URL and authentication headers.

Authentication

  • Authentication is provided at the MCP connection boundary.
  • Clients pass credentials in the MCP server configuration.
  • Systemiq validates access before exposing records, metrics, and related query surfaces.

Scope identifiers

  • The authenticated client boundary defines the highest query scope in the platform.
  • Use `system_id` for system scope within that client boundary.
  • Use `tool_id` for subsystem or processing scope within a system.
  • Use `element_id` for the most granular modeled entity.
  • Use `indicator` to address a stable output family.
  • Use `record_type` and `action` to narrow behavior and output shape.

Records

  • Records represent structured operational state and derived outputs.
  • They are typically used when an agent needs interpretable context objects rather than numerical analysis.
  • Common record dimensions include `system_id`, `tool_id`, `element_id`, `indicator`, `record_type`, `action`, and `created_at`.
  • Use records when you need scoped business context, agent outputs, or tool-generated insights.

Metrics

  • Metrics represent numerical values intended for aggregation, comparison, and time-series analysis.
  • The canonical metrics table is `metrics`.
  • Treat `created_at` as the primary timestamp for grouping and windowing.
  • Treat `value` as the numeric measure to aggregate.
  • Use `indicator` for the metric family and `key` for the metric variant, such as `value__latest`.
  • `system_id`, `tool_id`, and `element_id` may be `null` depending on where the metric was produced in the stack.

Query guidance

  • Prefer explicit scope by system, tool, element, or indicator.
  • Use records for structured context retrieval.
  • Use metrics for numerical and time-series analysis.
  • Query narrowed operational intent rather than broad raw enterprise state.
  • Discover indicators first before issuing narrower record or metrics queries.

Environment and access scope

  • Environment is derived from the authenticated token scope.
  • Clients do not need to pass `staging` or `production` manually through MCP queries.
  • Client scoping is also enforced by authenticated access rather than user-supplied parameters alone.

Core terms

  • Signal: raw operational evidence such as an event, file, API state, or business-system update.
  • Context: narrowed machine-consumable operational understanding derived from the model.
  • Operational memory: persisted records, metrics, summaries, and outputs available over time.
Stack modelConnectors & middleware