Are you an LLM? Read llms.txt for a summary of the docs, or llms-full.txt for the full context.
Skip to content
Architecture

Control Plane (State Runtime)

Use Mubit control APIs as the stable integration boundary for memory, context assembly, diagnostics, the learning loop, and coordination.

The control plane is the primary public integration boundary for current Mubit. It combines the memory lifecycle, compaction-safe context assembly, diagnostics, learning loop, coordination primitives, and structured run-scoped variables.

If your team needs one stable runtime boundary for long-running AI systems, standardize on the control plane first.

Core workflow model

WorkflowRecommended surfaces
Memory write/readclient.memory.remember, client.memory.recall, raw control.ingest, raw control.query
Context assemblyclient.memory.context / run.context, /v2/loop/context, /v2/control/context (lane="legacy")
Learning loopclient.lessons.reflect, client.outcomes.record / mubit.outcome, client.lessons.strategies
Diagnosticsclient.memory.health, client.memory.diagnose
Coordinationclient.agents.register, client.agents.handoffs.create, client.agents.handoffs.feedback

Operating guidance

  • Use one deterministic run_id strategy.
  • Prefer the global helpers and Client namespaces for application code.
  • Use client.checkpoint before compaction or risky transitions.
  • Use client.memory.context rather than rebuilding long prompts manually.

Failure modes and troubleshooting

SymptomRoot causeFix
Stale or weak responsesContext was built manually and inconsistentlyStandardize on client.memory.context()
The system does not improve across attemptsReflection or outcome recording is missingPair client.lessons.reflect() with client.outcomes.record()
Multi-agent behavior driftsRoles are implicitRegister agents and persist handoffs / feedback

Next steps