spec-challenge

Adversarial review for specifications is an independent critique process that challenges design plans to identify fatal flaws before coding begins. You need it when executing high-risk P0 or P1 cross-domain changes to catch critical design vulnerabilities early.

You can trigger automated adversarial review by manually invoking the /spec-challenge command on your target spec file. It also supports automatic triggering after plan generation completes for high-risk P0 and P1 cross-domain changes.

Yes, the adversarial review workflow enforces per-item user decisions on suggested improvements. After presenting a formal report with fatal flaws, you can choose to agree to modify, disagree, or mark items for discussion before batch applying improvements.

Yes, you can use adversarial spec review for high-risk cross-domain planning workflows. It automatically dispatches reviewer agents after plan generation completes for P0 and P1 cross-domain changes, ensuring structured critique before implementation.

Adversarial review reports and plan updates are persisted by saving artifacts to session data. The workflow updates session-state and plan files to reflect reviewer instructions, enabling full traceability and reproducibility of modifications across the lifecycle.