Source Trust Registry with Credibility Scores and Ingestion Policy Controls

Summary

Create a source governance layer that assigns trust metadata and credibility scores to each source, then uses that data during ingestion, ranking, and display.

Problem

The system aggregates many sources, but there is no centralized policy engine for source quality. Teams need a way to define which domains are trusted, conditional, or blocked, and to expose that trust level in search, article details, and newsletters.

Proposed Solution

• Add a Source Registry model for per-domain governance metadata.
• Calculate and store a sourceTrustScore and trustTier on ingested articles.
• Enforce policy actions during ingestion: allow, review, block.
• Surface trust badges in the frontend and expose trust filters in search.
• Weight newsletter ranking with trust scores (without completely hiding diverse viewpoints unless blocked).

Acceptance Criteria

• Registry supports fields like: domain, displayName, category, country, trustTier, policyAction, notes, updatedBy.
• Crawler/backend checks registry before persisting new articles.
• Blocked domains are skipped and logged with reason.
• Review-tier domains are ingested but marked needsSourceReview=true.
• Frontend shows trust badge on cards/details and allows trust-tier filtering.
• Admin API supports CRUD for source rules and bulk import/export.

Technical Notes

• Backend:
  • New collection: source_registry.
  • Update ingest pipeline to resolve source policy by normalized domain.
  • Expose endpoints such as:
    • GET /api/admin/sources
    • POST /api/admin/sources
    • PUT /api/admin/sources/:id
    • POST /api/admin/sources/import
• Frontend:
  • Admin management table for source rules.
  • Article UI trust badges and trust filter chips.
• Observability:
  • Metrics for blocked/reviewed/allowed ingest events.

Tasks

• Design source registry schema and domain normalization utility.
• Implement ingestion policy resolver (allow/review/block).
• Add article-level trust fields and migration/backfill for existing data.
• Build admin CRUD + import/export endpoints for source rules.
• Add frontend admin page and trust indicators on article views.
• Add ranking hooks to include trust weighting in feeds/newsletters.
• Add tests for policy enforcement and trust metadata propagation.
• Add docs for governance model and recommended default tiers.

Out of Scope (Initial Iteration)

• Fully automated fact-checking verdicts.
• Third-party reputation provider integrations.