Grading the Narrators: An Isnad-Rijal Framework for Claim-Level Provenance in Multi-Agent Knowledge Systems

Researchers adapt hadith scholarship concepts—especially isnad and rijal—to build a claim-level provenance and reliability framework for multi-agent AI systems.

The framework uses a graded narrator registry and chain-based decision logic to assess how trustworthy each transmission step is, not just the final answer.

On 20,000 physics-textbook claims, the tests support weakest-link quarantine and corroboration-based decisions, while some partial and unresolved failures remain.

The work could make multi-agent knowledge systems more accountable by tracking provenance and reliability across the full chain of transmission.