Context Continuation - Superwire

context is for conversation/message-history continuation.

agent investigate_task {
    model: model.fast
    instruction: "Investigate task {{ input.task_id }} and identify the main issue."

    output {
        issue: string
        evidence: [string]
    }
}

agent propose_solution {
    model: model.fast
    context: agent.investigate_task
    instruction: "Continue from the previous investigation and propose a solution."

    output {
        solution: string
        steps: [string]
    }
}

context: agent.investigate_task passes the whole message history of investigate_task into propose_solution. The second agent then appends its own instruction and continues the work.

Context versus structured output

Use a structured reference when you need a value:

{{ agent.investigate_task.issue }}

Use context: agent.name when you want the next agent to continue the prior conversation:

context: agent.investigate_task

These are different operations. Structured references read JSON-like output fields. Context continuation passes message history.