Use Case

As a Node.js driver user building streaming pipelines (aggregation cursor → stream → batch writers) with progress logging
I want AggregationCursor.clone() to represent an independent cursor for a variant of the same query (e.g. append $count on the clone, iterate documents on the original)
So that I can obtain a total count and then stream the same logical result set without mutating the original cursor’s pipeline or result shape
 

User Experience

Expected outcome after fix or doc change:

• Pattern cursor.clone().addStage({ $count: '…' }).next() for count, then cursor.stream() / toArray() on the original cursor, yields the same documents as if $count had never been added to the pipeline used for iteration.
• Alternatively, if sharing the pipeline is intentional: JSDoc on clone() and addStage() clearly states that both cursors share the same pipeline array and that addStage on either cursor affects both.

If treated as bug

• Impacted customers: Any user of the official driver who combines clone(), addStage() (or project / match / etc.), and a second use of the original cursor.
• Severity: Functional / data correctness — not a crash. Easy to misdiagnose as application or MongoDB query logic.
• Blocked / broken: Not a hard blocker for the whole ecosystem; workaround is two aggregate([...]).

Dependencies

• None

Risks/Unknowns

• Behavior change: Copying the pipeline in clone() ([...this.pipeline] or shallow copy of stage objects) could break rare code that intentionally mutates cursor.pipeline on one cursor expecting the other to see it (unlikely but possible).

Acceptance Criteria

Implementation Requirements

• Option A (preferred for UX): AggregationCursor.clone() passes a copy of the pipeline array to the new cursor (e.g. [...this.pipeline]), so subsequent addStage on the clone does not alter the original cursor’s pipeline (and vice versa, if addStage is allowed on uninitialized originals only).
• Option B (minimum): No code change; document on clone() that pipeline is shared and that count + stream requires separate aggregate() calls or [...stages].

Testing Requirements

• Unit/integration test: fixed pipeline with $project → clone().addStage({ $count: 'count' }).toArray() → assert original toArray() / iteration still returns projected docs, count matches length (or $count value).
  I have only added a unit test on pipeline field

Documentation Requirements

• JSDoc on clone() (and optionally addStage()): shared or copied pipeline.

Follow Up Requirements

• If Node copies pipeline but other drivers share: file follow-up for cross-driver consistency or document Node-specific behavior in driver comparison docs.