Expand description
Core types for multi-agent coordination.
Structs§
- Agent
Output - Output from one agent’s execution.
- Agent
Spec - Blueprint for an agent in a multi-agent system.
- Message
- A message between agents in a multi-agent system.
- Token
Accounting - Token and cost accounting contributed by an AgentRunner. Runners opt in by
setting
AgentOutput::tokens; benchmarks that need economic metrics read it.