challenge

challenge — adversarial decision review

This is deliberately user-invoked. A permanently automatic objection reflex would distort the user's decision process and make every ordinary preparation task defensive.

Goal

Attack one user-approved strategy or decision plan and reduce the result to one root objection plus one cheap falsification test. The user still owns the decision.

Workflow

1. Identify the exact decision and the assumptions it needs.
2. Check the strongest failure axes: false facts, unsupported inference, missing evidence, hidden trade-off, unsafe scope, or a result assumed before it exists.
3. Collapse related findings into one load-bearing root.
4. Name the cheapest test that could show the root matters or does not matter.
5. Return the objection and test for user review before the plan continues.

Rules

• Run only after explicit user opt-in or an explicitly user-confirmed consequential-decision policy.
• Do not produce a total, hiring prediction, recommendation, or decision on the user's behalf.
• Do not average independent axes or hide a confirmed hard conflict.
• One root and one test; no checklist expansion.
• This Skill is read-only and never invokes another Skill.

Gate D terminal semantics

• No meaningful load-bearing objection survives → report completed.
• A load-bearing objection is found → report needs_approval with the {root, first_nail} summary; the plan must pause for the user's review before any dependent Quality Skill runs.
• Do not report an objection as completed merely because the analysis itself finished.

Verification

Report {root, first_nail} with the evidence behind the objection. If no load-bearing objection survives, say so and name the assumptions that remain Unknown.

Adaptation notice: ../../_shared/THIRD_PARTY_NOTICES.md.