KAVACHKAVACH
v2.0.1

Reviewing Governance Decisions

Inspect a persisted decision from outcome to evidence, explanation, lineage, and audit context.

Goal

This tutorial walks through a read-only review of a governance decision in Kavach Studio. You will start with the decision outcome, validate the evidence used to reach it, and trace the result through the ontology graph.

Prepare the workspace

Start the local stack and seed the representative ontology and decision data:

bash
docker compose up --build
curl -X POST http://localhost:8000/api/v1/ontology/demo/seed

Open Studio at http://localhost:3000/decisions. The demo seed includes approved, proposed-for-review, rejected, and blocked decisions.

Review the decision

  1. Select a decision from the index, or paste a decision ID into Decision Lookup.
  2. Start with status, target, decision type, confidence, and policy outcomes.
  3. Read the explanation timeline to understand the deterministic reasoning sequence.
  4. Open the Evidence Graph and inspect the supporting facts and evaluation results.
  5. Open Decision Lineage and follow relationships to the target, policies, actor, jobs, and related assets.
  6. Finish by checking audit records, request ID, and correlation ID.

Interpret the graphs

The Evidence Graph shows the bounded reasoning context used by the decision. The Decision Lineage view shows the ontology projection around the decision. They overlap intentionally, but they are not interchangeable: evidence supports the conclusion, while lineage explains governance relationships and provenance.

Use the API

Studio uses these read endpoints for the inspection view:

http
GET /api/v1/decisions
GET /api/v1/decisions/{decision_id}/detail
GET /api/v1/decisions/{decision_id}/evidence
GET /api/v1/decisions/{decision_id}/explanation
GET /api/v1/decisions/{decision_id}/lineage?depth=2

In a multi-tenant deployment, send the organization and project scope selected in Studio. The decision repository and ontology graph are separate persistence surfaces, so verify both when diagnosing missing data.

Review checklist

  • Is the decision target the expected asset and version?
  • Are the policy outcomes and confidence consistent with the evidence?
  • Can each important conclusion be traced through the evidence graph?
  • Does lineage identify the governing policy, actor, and source assets?
  • Are request, correlation, and audit records sufficient to reproduce the investigation?