Multi-Agent Coordination Monitoring
Track how agents communicate, delegate tasks, and coordinate. Detect conflicting instructions, circular delegations, and unexpected agent appearances.
Python SDK
from prysmai.detectors import MultiAgentMonitor
monitor = MultiAgentMonitor(
expected_agents=["planner", "researcher", "writer"],
max_delegation_depth=3,
detect_conflicts=True,
)
session.attach_detector(monitor)
Detection Types
| Type | Description |
|---|---|
| unexpected_agent | An agent not in the expected list appeared |
| circular_delegation | Agent A → B → C → A delegation cycle detected |
| deep_delegation | Delegation chain exceeded max_delegation_depth |
| instruction_conflict | Two agents received contradictory instructions |
| orphaned_delegation | A delegation was sent but never received |
Agent Network View
The Agent Network tab in the dashboard shows:
- All tracked agents with event counts and delegation metrics
- Delegation flows between agents with direction and frequency
- Real-time topology of your multi-agent system
Dashboard
Navigate to Dashboard → Multi-Agent to view the overview, event log, and agent network.