AgentDefinition
Defined in: scoring/agents.ts:84
Complete agent definition
Type Parameters
Section titled “Type Parameters”T = unknown
Properties
Section titled “Properties”capabilities
Section titled “capabilities”capabilities:
AgentCapabilities
Defined in: scoring/agents.ts:100
What this agent can do
cost:
number
Defined in: scoring/agents.ts:93
Cost per invocation in your chosen units Could be cents, tokens, or relative units (0=free, 100=expensive)
enabled
Section titled “enabled”enabled:
boolean
Defined in: scoring/agents.ts:102
Whether this agent is currently enabled
execute
Section titled “execute”execute:
AgentExecutor<T>
Defined in: scoring/agents.ts:109
The executor function - implement this for your provider
id:
string
Defined in: scoring/agents.ts:86
Unique identifier (e.g., ‘ollama-qwen’, ‘jules’, ‘openai-gpt4’)
name:
string
Defined in: scoring/agents.ts:88
Human-readable name
priority
Section titled “priority”priority:
number
Defined in: scoring/agents.ts:98
Priority when multiple agents can handle same tier Lower = preferred (will be tried first)
requiresApproval?
Section titled “requiresApproval?”
optionalrequiresApproval:boolean
Defined in: scoring/agents.ts:107
Require explicit approval before using? Useful for expensive agents - task.metadata.approved must include this agent’s ID