At a glance
| Input | Action | Output | Limit |
|---|---|---|---|
| Approved cyclic edge | Detect a previously visited node | Recorded cycle entry | Candidate and rejected edges remain excluded |
| Forward direction | Use forward adjacency links | Bounded downstream result | Direction changes which links are followed |
| Reverse direction | Use reverse adjacency links | Bounded provenance result | It does not infer an inverse relationship |
| Hop and result settings | Stop at enforced bounds | Explicit truncation state | Up to 20 hops and 1,000 results per request |
How it works
- Open the retained graph in Enterprise evidence workspace.
- Select Forward impact or Reverse provenance.
- Review the returned nodes and links.
- Inspect cycle entries in the result evidence.
- Check hop and result truncation states.
- Correct any relationship that should not be circular.
Example
Requirement A derives Requirement B, while another approved edge links B back to A. Traversal records the return edge as a cycle. It keeps both requirements visible and does not revisit A indefinitely. The team can then decide whether the circular relationship is valid or a governance error.
What this does not mean
Cycle handling protects query execution. It does not decide whether a circular relationship is correct. Evidence owners remain responsible for reviewing the graph and correcting invalid links.
Related questions
Next step
Review every recorded cycle before relying on the trace for a release or investigation.
View on Atlassian Marketplace