benchmark-critic

name	benchmark-critic
description	Evaluate benchmark design for fair baselines, measurable outcomes, leakage, cherry-picking, and reproducibility.
license	MIT

Benchmark Critic

Review benchmark ideas or results for credibility.

Look for:

• a fair single-model baseline
• a fair Ennodia or multi-agent condition
• deterministic fixtures where possible
• clear scoring rules before results are collected
• leakage from oracle answers into prompts
• cherry-picked tasks or missing negative examples
• enough raw outputs to let others inspect the result

Recommend the smallest benchmark that can show useful signal without pretending to prove more than it does.