orchestrator-first

What problem does it solve?

Orchestrator-first mode centralizes control to the main agent, who owns scope, dispatch, integration, validation, and final synthesis, while subagents implement dedicated lanes. This separation simplifies governance and scaling for complex, multi-agent conversations.

Core Features & Use Cases

Prereqs: User explicitly enables orchestrator-first mode; the canonical prompt source exists at references/prompts/orchestrator-first.md; subagent delegation is available and allowed by active runtime instructions before any subagents are spawned. Inputs: Optional goal, constraints, lane boundaries, validation expectations, or disable instruction; Future user requests in the same thread when the mode has been explicitly enabled. Outputs: Confirmation that orchestrator-first mode is enabled when invoked as a mode-setting prompt; Delegation gate and disable instruction from the canonical prompt source; No subagents unless the current delegation gate allows and runtime rules permit it. Exit codes: N/A (conversation workflow; no repo scripts). Failure modes: Prompt source is missing or empty; The request is too small, unclear, tightly coupled, destructive, or blocked on subagent output; Active runtime instructions prohibit delegation despite the requested mode.

Quick Start

Enable orchestrator-first mode for this conversation thread.