Configure intentional silence

What you will build

A predictable abstention path that explains why no answer was produced and what evidence was missing.

Inputs and ownership

• Bounded use case
• Required evidence definition
• Approved response language

Blueprint

This illustrates the governed fields to define during onboarding; it is not a public production API contract.

{
  "required_basis": ["active_source", "claim_support"],
  "on_missing_basis": {
    "action": "abstain",
    "reason_visible": true
  }
}

Build it step by step

1. 01

   Define minimum evidence

   List the source types or claim support required before the system may answer.

2. 02

   Choose the silence response

   Use language that is clear to users and does not imply a hidden answer exists.

3. 03

   Test edge cases

   Run partial, expired and conflicting evidence scenarios.

4. 04

   Audit abstentions

   Confirm that the missing basis and applied policy are visible in the trace.

Prove the governed path

• The permitted path produces an answer from approved evidence.
• An unsupported request returns an explicit governed outcome.
• Every material claim resolves to a known knowledge unit.
• The decision can be rerun after a source version changes.