hiadrianchen

challenge-plans

Before you execute a drafted plan/spec/design doc, run a multi-agent adversarial cross-review to surface the flaws that cause downstream rework, aggregating "evidenced, cross-family-verified" objections into a verdict. Use when the user asks to "review this plan/spec", "can this approach be executed", "poke holes / adversarial review / QA this", "harden before executing", or when an agent is about to hand a drafted decision/QA back to the user — first run this skill and present the cross-review recommendation plus surviving objections. Runs on local subscription CLIs (claude/codex), no per-token API cost. Not for "help me pick among options" — that's the weigh-options deliberation skill.