Plain definitions for the terms behind BadMars — adversarial analysis, red-teaming, pre-mortems, and the words we coined for the work. Each entry opens with a one-sentence definition, then how it applies to a plan.
AI built to attack an input rather than assist with it — to find where a plan fails.
Read the definition →Attacking your own plan from an opponent's point of view to surface what would break it.
Read the definition →Imagining the plan has already failed, then working backwards to name the causes.
Read the definition →Testing whether a plan's core assumptions survive contact with evidence, customers, and competitors.
Read the definition →A specific way a plan can break — a wrong assumption, a missing number, a step built on something untrue.
Read the definition →The bench of adversarial specialist personas — up to thirteen — that convenes on your plan.
Read the definition →The single biggest problem to fix first — the one that quietly sinks the rest.
Read the definition →The filed output: Score, Verdict, ranked Findings, and the Frog — web, PDF, and a share card.
Read the definition →The free partial read (Flyby) and the full-Panel paid report (Landing).
Read the definition →