ralplan

To create an auditable plan for an underspecified software task, use a consensus planning loop with Planner, Architect, and Critic roles to iteratively define scope, decision drivers, and acceptance criteria before execution begins.

Multi-agent consensus planning for high-risk software engineering is a governance workflow where Planner, Architect, and Critic agents collaborate to transform vague requests into concrete plans with documented risks, tradeoffs, and ADRs.

You can enforce risk management and scope definition before execution by applying a gatekeeper loop that requires explicit acceptance criteria, test plans, and ADR documentation to prevent unbounded work on underspecified requests.

Yes, multi-agent planning supports interactive prompts and deliberate modes to facilitate team alignment, ensuring decision drivers and viable options with tradeoffs are explicitly defined and documented through ADRs.

You should use a consensus planning loop instead of direct execution when facing high-risk or underspecified tasks, ensuring an architect-critic mechanism defines scope and acceptance criteria to prevent unbounded execution.