Skip to content

Bound the deterministic-plus-LLM prioritization contract #7

Description

@BleedingDev

Question

How should deterministic evidence scoring, structured LLM reasoning and reranking, and deterministic post-validation combine into one explainable priority list? Define the separate scores, LLM adjustment bounds, evidence-citation requirements, fallback when no model is available, audience-language generation, and guards against invented impact or unstable ordering.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions