What problem does it solve? MongoDB applications fail in predictable ways: unbounded embedded arrays that hit the 16 MB document limit, indexes ordered wrong for the query shape, aggregation pipelines that scan millions of documents, and writes lost or duplicated during failovers. This Skill gives an AI assistant the measured, version-gated rules to diagnose and fix these problems instead of guessing. ## Core Features & Use Cases - Document modeling and migration: Decide embed-versus-reference from access patterns, bound every array, split hot and cold fields, and run online schema migrations with versioned documents and staged $jsonSchema validation. - Query and index tuning: Read explain() counters (totalKeysExamined, totalDocsExamined, nReturned), design compound indexes in Equality-Sort-Range order, and remove indexes safely with hideIndex before dropIndex. - Aggregation and durability: Fix $lookup fan-out, blocking-stage spills, and unbounded $group accumulators; configure retryable writes, write concern, and withTransaction so writes survive elections. - Use Case: A team's chat backend slows as threads grow. The Skill identifies the unbounded messages array as the root cause, prescribes a child collection keyed by thread, fixes the sidebar's missing projection, and lays out a batched online migration. ## Quick Start Ask the assistant to review your MongoDB data model, slow query, or aggregation pipeline using the mongodb skill and paste the relevant code or explain output.