hermes-rubric
by Community
Evidence-first LLM-as-judge scoring for AI artifacts — papers, PRs, prompts, cold emails: synthesizes a rubric, collects quoted-evidence citations, scores only against that evidenc
OSS
hermes-rubric
Added 1 Oct 2026
Overview
Evidence-first LLM-as-judge scoring for AI artifacts such as papers, PRs, prompts, and cold emails. It synthesizes a rubric, collects quoted-evidence citations, scores only against that evidence, and hedges on thin evidence. Each dimension ties to a file:line or quote, with reproducibility receipts, and supports 7 backends.
Best for
Best for
Developers who need auditable, evidence-cited LLM scoring for code, prompts, or written artifacts
Use cases
- Score pull requests against an evidence-backed rubric
- Evaluate prompt quality with quoted citations
- Audit AI-generated text with reproducible file:line references
Notes
2 stars on GitHub. Last updated 2026-09-29. Licensed Apache-2.0.
Pros
- Grounds scores in quoted evidence rather than model intuition
- Provides reproducibility receipts with file:line or quote references
- Supports 7 backends for deployment flexibility
Cons
- Early-stage project with few GitHub stars
- Python-only implementation limits integration options
- Requires collecting evidence citations before scoring
Get the free Developer’s Field Guide
A 27-page field guide to the AI coding workflow with Claude. Claude Code, MCP servers, the prompt patterns that work, and what to delegate. Free.
Enter your work email. We send it straight over, plus a few short notes worth knowing. Unsubscribe any time.