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
| Workflow | Recommended surfaces |
|---|---|
| Memory write/read | client.memory.remember, client.memory.recall, raw control.ingest, raw control.query |
| Context assembly | client.memory.context / run.context, /v2/loop/context, /v2/control/context (lane="legacy") |
| Learning loop | client.lessons.reflect, client.outcomes.record / mubit.outcome, client.lessons.strategies |
| Diagnostics | client.memory.health, client.memory.diagnose |
| Coordination | client.agents.register, client.agents.handoffs.create, client.agents.handoffs.feedback |
Operating guidance
- Use one deterministic
run_idstrategy. - Prefer the global helpers and
Clientnamespaces for application code. - Use
client.checkpointbefore compaction or risky transitions. - Use
client.memory.contextrather than rebuilding long prompts manually.
Failure modes and troubleshooting
| Symptom | Root cause | Fix |
|---|---|---|
| Stale or weak responses | Context was built manually and inconsistently | Standardize on client.memory.context() |
| The system does not improve across attempts | Reflection or outcome recording is missing | Pair client.lessons.reflect() with client.outcomes.record() |
| Multi-agent behavior drifts | Roles are implicit | Register agents and persist handoffs / feedback |
Next steps
- Review the full route contract at Control HTTP reference.
- Apply the workflow in Support agent memory loop.