team-lifecycle-v4

Multi-agent software development orchestration routes work through analyst, writer, planner, executor, tester, reviewer, and supervisor roles using a dependency DAG. This coordinates the complete lifecycle from specifications to code changes, testing, and reviews while maintaining structured dependencies and supervised checkpoints.

Checkpoint supervision for spec and plan consistency uses a resident supervisor agent to validate alignment between specification artifacts and implementation plans. It blocks or warns execution when quality thresholds are not met, ensuring structured dependencies are respected throughout the workflow.

To start an end-to-end multi-agent development workflow, provide a clear task description and optionally specify a desired role. The orchestration routes work through analyst, planner, executor, tester, and reviewer roles, managing session artifacts and checkpoint verdicts automatically.

Multi-agent workflow orchestration supports role routing via the --role parameter, allowing you to designate specific agents like writer, planner, or executor for targeted tasks. This ensures dependency-respecting execution across the full software development lifecycle.

You can reuse findings from earlier multi-agent development sessions through sessionized collaboration and knowledge transfer. The workflow maintains a session directory containing artifacts, wisdom, and an exploration cache so later tasks can build upon earlier discoveries.