review-all-gdds

Run a cross-GDD review that loads every system GDD and compares dependency sections, rules, tuning knobs, formulas, and acceptance criteria pairwise. Contradictions are flagged as blocking issues or warnings with citations to the exact documents and sections involved.

A single GDD review checks one document for internal completeness, while a cross-GDD review examines the relationships between all documents. Cross-GDD review catches stale references, dependency asymmetries, and design issues like dominant strategies that only appear when systems are viewed together.

Run it after all MVP-tier GDDs are individually approved, after any GDD is significantly revised mid-production, and before architecture work begins. Architecture built on inconsistent GDDs inherits those inconsistencies.

Yes, the since-last-review mode uses git history to identify GDDs modified since the last review report, then scopes the full read to those documents plus their listed dependencies. This avoids re-reading unchanged documents.

The review stops and reports that cross-GDD analysis requires at least two system GDDs. You must write additional GDDs first, then re-run the review.

It detects competing progression loops, cognitive overload from too many simultaneous active systems, dominant strategies, economic source/sink imbalances, mismatched difficulty scaling curves, pillar drift, anti-pillar violations, and conflicting player fantasies.