# Risk explainability

Purpose
- Provide per-factor contributions with provenance and gating rationale.

Explainability envelope
- job_id, tenant_id, context_id
- profile_id, profile_version, profile_hash
- finding_id, raw_score, normalized_score, severity
- signal_values and signal_contributions
- override_applied, override_reason, gates_triggered
- scored_at and provenance hashes

UI and CLI expectations
- Deterministic ordering by factor type, source, then timestamp.
- Highlight top contributors and gates.
- Export Center bundles include explain payload and manifest hashes.

Determinism and offline
- Fixtures under docs/risk/samples/explain/ with SHA256SUMS.
- No live calls in examples or captures.

Related references
- risk/overview.md
- risk/factors.md
- risk/formulas.md
- risk/profiles.md
- risk/api.md