Skip to content
Junjo AI

AgentInvocation

junjo.evaluation.AgentInvocation

Kind: Class

SDK version: 0.69.0

Documentation channel: Stable release

Signature

Section titled “Signature”
AgentInvocation(agent: Agent[AgentInputT, AgentOutputT, DependenciesT, AgentStateT, StoreT], input: AgentInputT, dependencies: DependenciesT, store: StoreT | None = None, history: tuple[AgentMessage, ...] = (), correlation: ExecutionCorrelation | None = None, cleanup: Cleanup | None = None)

View source

Agent definition and application-owned invocation values for one case.

Members

Section titled “Members”

agent

Section titled “agent”
agent: Agent[AgentInputT, AgentOutputT, DependenciesT, AgentStateT, StoreT]

View source

Agent definition used to execute this case with its chosen model and tools.

input

Section titled “input”
input: AgentInputT

View source

Typed application input supplied to the Agent.

dependencies

Section titled “dependencies”
dependencies: DependenciesT

View source

Application services available to this Agent execution and its tools.

store

Section titled “store”
store: StoreT | None = None

View source

Optional borrowed application Store; omitted uses the Agent factory.

history

Section titled “history”
history: tuple[AgentMessage, ...] = ()

View source

Conversation history supplied by the application for this case.

correlation

Section titled “correlation”
correlation: ExecutionCorrelation | None = None

View source

Optional trusted application correlation identity propagated through this execution.

cleanup

Section titled “cleanup”
cleanup: Cleanup | None = None

View source

Optional per-case cleanup, called after execution or failure, including projection failures.