Awesome Reviewers

Verification · Finding validation

Finding deduplication

Merges findings that describe the same problem from different reviewers, files or runs into one.

Sub-techniques

  • Semantic dedup V1045.001

    Cluster findings by embedding similarity and location.

  • Consolidation V1045.002

    Merge a cluster into one finding that keeps all evidence.

  • Fingerprinting across runs V1045.003

    Stable fingerprints so a finding is recognised on the next commit.

How to build it

  1. Emit findings as SARIF or a common schema with file, line and rule.
  2. Fingerprint by rule, file and normalised code, not line number.
  3. Cluster the remainder by embedding and merge.

Signals in a repository

  • SARIF partialFingerprints, or a findings table keyed by fingerprint.