AgentStack
SKILL verified MIT Self-run

Paper Critic Zh

skill-gy910210-hermes-research-skills-paper-critic-zh · by gy910210

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。

No reviews yet
0 installs
0 views
view→install

Install

$ agentstack add skill-gy910210-hermes-research-skills-paper-critic-zh

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Paper Critic Zh? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Hermes 适配说明

  • 本 skill 现面向 Hermes 使用,优先依赖 Hermes 原生工具:search_filesread_filewrite_filepatchsession_searchdelegate_taskcronjobbrowserweb/searchvision
  • 若正文提到 references/...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。

Paper Critic(中文)

输入

  • paper_notes(推荐)
  • evidence_map(可选)
  • existing_claims 或拟写结论(可选)

输出

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

步骤

  1. 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
  2. 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
  1. 对 benchmark、release、私有数据、license、未披露规模等信息做事实性审校。
  2. 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。
  3. 对每个子方向产出:
  • strongest_support
  • strongest_contradiction
  • still-missing evidence
  • risky wording to avoid
  1. 给出可直接用于改写主稿的 rewrite_suggestions

守护

  • 所有审校意见都要能回指到原始来源。
  • 审校器不是为了“挑刺”,而是为了把结论写稳。
  • 不能确认的事实应降级为 未披露待验证
  • 对产业侧信息,优先区分“官方页面可确认”“高质量报道可观察”“内部说法不可证实”三层。

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.