Skip to main content

Evidence IDs

Evidence IDs are the glue that ties rendered findings back to the machine-native artifacts that produced them.

What the example output shows​

The example bundle surfaces evidence IDs in several places:

  • raw module outputs often include module-scoped evidence ID fields
  • fast-fail risk signals include evidence_id
  • the wet-lab readiness machine artifact includes an artifact.evidence_id
  • paths inside failure-mode evidence records can point back into orchestrator outputs

Example IDs use the EID-... pattern and appear alongside specific risk or artifact records rather than only in one central index.

Why evidence IDs matter​

An evidence ID should let an operator or auditor answer:

  • which module produced this signal
  • which artifact should be opened next
  • whether the claim is traceable to emitted machine output

Without evidence IDs, narrative findings become much harder to audit.

How evidence travels​

In the current output model, evidence can move through several layers:

  1. raw module artifact
  2. combined or orchestrated result
  3. fast-fail or failure-mode render layer
  4. report narrative or recommendation layer

Evidence IDs make those hops inspectable.

Whenever a result page surfaces a specific finding, document:

  • the artifact where the signal originates
  • the evidence ID, if emitted
  • the field path used in render or aggregation layers

That keeps the report linked to the machine-native record.