steps/agents

Low-level agent and model execution functions for factory-owned step wrappers.

Workflow code normally uses runAgent, askAgent, askModel and askJev from #jigs/routines. These implementations must run inside "use step" code. See Models and harnesses.

Execution primitives

executeAgent()

function executeAgent(wire, metadata): Promise<
  | AgentResult
  | {
  jitFailure: object & object[];
}
  | {
  resumeFailed: string;
}>;

Run or ask an agent harness, checking worktree requirements before a run.

Parameters

Parameter	Type
wire	AgentRequest
metadata	RunMetadata

Returns

Promise< | AgentResult | { jitFailure: object & object[]; } | { resumeFailed: string; }>

---

executeJev()

function executeJev<QUESTIONS>(wire, metadata): Promise<JevResult<QUESTIONS>>;

Evaluate typed questions with a decision-capable model.

Type Parameters

Type Parameter
QUESTIONS extends JevQuestions

Parameters

Parameter	Type
wire	AskJevOptions<QUESTIONS>
metadata	RunMetadata

Returns

Promise<JevResult<QUESTIONS>>

---

executeModel()

function executeModel(wire, metadata): Promise<ModelResult>;

Ask an API-backed model source.

Parameters

Parameter	Type
wire	ModelRequest
metadata	RunMetadata

Returns

Promise<ModelResult>