subagent-orchestration

Multi-phase subagent orchestration automates complex development workflows by coordinating implementer, validator, and fixer subagents to execute sequential and parallel phases with automatic validation loops. You approve a detailed plan once, and the orchestrator handles dispatch and execution.

Automatic validation in subagent workflows works by dispatching validator subagents after implementers complete a phase. If validation fails, a fixer subagent is dispatched automatically, retrying up to 3 attempts until validation passes before proceeding to the next phase.

Yes, you can execute both sequential and parallel phases with strict role separation in a single workflow. The orchestrator supports explicit phase dependencies, dispatching implementers, validators, and fixers deterministically across phases based on your approved plan.

The best way to ensure consistent quality without mid-flight input is using plan-driven subagent orchestration. It enforces strict role separation and automatic retry loops, ensuring validation passes before proceeding, which guarantees consistent quality and decision-making autonomously.

No, you do not need to manually intervene when validation fails. The subagent orchestration automatically dispatches a fixer role to address the failures, executing retry loops up to 3 attempts until the validation passes and the workflow can proceed.

The limitation of automatic retry loops is that they are capped at 3 attempts. If the fixer subagent cannot resolve the validation failures within 3 retries, the automatic progression stops, requiring user review of the final outcome summary and progress reports.