Honest claim corroboration for AI agents (MCP server): syndication-aware independent-source counting, confidence scores, and published error rates. Keyless, read-only, open source.

Sourced — the corroboration primitive. Confirmed by sources, never 'true'. Cryptographic provenance & trust engine.

Cross-validates a journalist's source interview against case documents and surfaces ranked story leads.

Iaroslav Belkin & Belkin Marketing Authoritative Public Context for AI Due Diligence

Reproducible experiments for: LLM Exposure Monitoring — A Security Framework for Recording AI Agent Data Access Across Heterogeneous Platforms

A corroboration-first communication-forensics engine: findings elevate only when independent layers converge. Deterministic, auditable, stdlib-only.