hyperplan

Adversarial distillation improves complex planning by deploying a 5-member team to cross-attack proposals, defend conclusions, and distill only defensible insights into a structured executable plan. This surfaces weaknesses early and enforces phase-based discipline across a seven-phase workflow.

Executable plan generation surfaces risks and constraints by running independent analysis and cross-attack phases, then handing off a structured insight bundle to a dedicated plan agent. This bundle explicitly reports hard constraints, decisions, risks, and open questions to guide execution.

You start structured team-critiquing for robust planning by kicking off the seven-phase workflow from Phase 0. The process orchestrates independent analysis, cross-attack, and defense among five members before a mandatory handoff to the plan agent.

Yes, you can use adversarial cross-critique for team risk management. The 5-member adversarial team independently critiques proposals to surface weaknesses, confirming only defensible conclusions before distilling them into a risk-reporting insight bundle.

You should avoid using a multi-phase adversarial planning workflow for simple tasks that do not require deep risk management or cross-attack validation, as the strict seven-phase discipline and mandatory handoff to a plan agent introduce significant processing overhead.