Investigate AI-agent analysis report on the crud specification. This ticket tracks findings identified during a systematic review of the spec: missing test coverage, ambiguous normative language, and spec/test inconsistencies.

Missing Tests

• countDocuments with skip > collection size — behavior with oversized skip not tested
• distinct on a field that doesn't exist in any document — should return empty array
• aggregate with $out/$merge + bypassDocumentValidation interaction
• Bulk write splitting at exactly 2 MiB boundary — spec defines size limits but exact boundary behavior is not
  tested
• CSOT interaction with tailable cursor await timeouts
• estimatedDocumentCount with comment on servers < 4.4 — behavior when server doesn't support the option
• rawData option for non-time-series collections — error handling when sent to pre-8.2 servers or wrong collection
  type

Ambiguities

• bypassDocumentValidation without $out/$merge: Spec says "only applies when $out or $merge stage is
  specified" — should an explicit bypassDocumentValidation=true on a pipeline without these stages error or be
  silently ignored?
• comment + hint interaction: Spec mentions comment can be any BSON type (4.4+) but doesn't specify if comment
  affects query planning alongside hint.
• rawData documentation scope: Marked "internal use by MongoDB teams" but criteria for when to include it in
  public vs. internal documentation is vague.

Inconsistencies

• countDocuments vs. estimatedDocumentCount options: countDocuments has skip/limit/hint;
  estimatedDocumentCount explicitly does not — rationale not explained in spec.
• batchSize change between operations: "SHOULD apply to both original aggregate command and subsequent getMore" —
  doesn't clarify if batch size can change between operations mid-cursor.
• comment field server version requirements: Different operations have different version thresholds for comment
  support; version matrix is scattered across sections rather than centralized.

Notes

• ~378 test files. Strong happy-path coverage across all CRUD operations. Collation, hint, and comment interactions in
  complex queries lack depth.
• Server version compatibility for newer options (rawData, comment as BSON type) not comprehensively covered.
• Edge cases around cursor timeouts and bulk write splitting are under-tested.