explore-critique-implement

Adversarial critique for uncertain problem spaces is coordinated by concurrently exploring options via dedicated agents, validating them through separate critic roles, and implementing refinements in iterative cycles until maturity is reached.

Validating multiple plausible approaches requires a persistent explorer to surface options, followed by adversarial critique from separated roles to reduce bias, and iterative refinement loops to ensure traceable and auditable decisions.

Iterative exploration with traceable decisions is implemented by maintaining a project-understanding ledger, enforcing structured agent roles, and handling explicit blockers during the explore, critique, and implement loop.

Iterative exploration with adversarial critique requires no external dependencies or environment setup, relying entirely on internal agent coordination to manage the persistent context and structured handoffs.

An explore-critique-implement cycle should be used for non-trivial problems with uncertain solution spaces where multiple plausible approaches must be weighed, validated, and refined through persistent context and adversarial critique.