Paper critic zh

Research Skills for Hermes Agents

Install

npx -y skills add gy910210/hermes-research-skills --skill paper-critic-zh

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

Copied from the file, not written here

证据审校器：基于 paper_notes 或来源包审查直接证据、综合判断、过度推理、benchmark/release/private data/license 风险，并生成中文 evidence_audit。需要把研究内容从“能写”提升到“写得稳”时调用。

2.3 KB, as published. Nobody here has run it

本 skill 现面向 Hermes 使用，优先依赖 Hermes 原生工具：search_files、read_file、write_file、patch、session_search、delegate_task、cronjob、browser、web/search、vision。
若正文提到 references/... 或 scripts/...，优先读取当前 skill 目录下对应文件，不再依赖 Claude 专属目录结构。
原始 Claude agents/openai.yaml 不作为执行前提；需要并行研究、分工精读或角色评审时，改用 Hermes 的 delegate_task。
保留原有研究方法论与产物契约，但执行层统一按 Hermes 工具体系落地。

evidence_audit
- 字段契约见 ../research-orchestrator-zh/references/artifact-contracts.md

对每条拟写结论判断它属于 直接证据、邻近证据 还是 综合判断。
检查是否存在常见过度推理：
- 从 offline 指标推出 online gain
- 从模拟器推出真实用户迁移
- 从 demo 推出可部署
- 从行业报道或产品观察推出平台内部结论
- 从单个入口形态推出底层技术路线已经成立
对 benchmark、release、私有数据、license、未披露规模等信息做事实性审校。
若来源包含行业报道、采访、产品观察或二手传闻，补充一层 publicly_verifiable / industry_observation / rumor_or_unverified 标注。
对每个子方向产出：
- strongest_support
- strongest_contradiction
- still-missing evidence
- risky wording to avoid
给出可直接用于改写主稿的 rewrite_suggestions。