Skip to content
Agentmart
Submit a tool

hypo-review

Run a multi-persona simulated academic review for a LaTeX or PDF paper.

Skill by HypoxanthineOvO

IndexedMIT★ 2
Write a review

Chalk Labs

Need help deploying hypo-review?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/hypoxanthineovo-hypo-review

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Run a multi-persona simulated academic review for a LaTeX or PDF paper. Parses paper structure, prepares reviewer-specific prompts, then guides independent reviews, AC meta-review, revision roadmap, and consistency checks.

Use cases: documents, research

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/hypo-review

What's inside

Skill name
hypo-review
Description
Run a multi-persona simulated academic review for a LaTeX or PDF paper. Parses paper structure, prepares reviewer-specific prompts, then guides independent reviews, AC meta-review, revision roadmap, and consistency checks.
License
MIT
Instructions
129 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# /hypo-review — 多角色模拟审稿

你是一个学术论文模拟审稿系统的控制器。目标是让 Agent 依次扮演多个虚拟审稿人,对同一篇论文给出独立、结构化、可执行的审稿意见。

## 参数

$ARGUMENTS
- paper_path: 论文路径,支持 `.tex` / `.pdf`
- venue: 目标会议或期刊,可选,如 `dac`, `neurips`, `tcas1`
- panel: 审稿团,`default` / `full`
- reviewers: 自选审稿人 id,可选
- severity: 苛刻程度,`gentle` / `standard` / `harsh`
- domain: 论文领域,可选;用于覆盖自动推断

## 工作流程

### 第零步:文献检索(默认开启)

自动从 Semantic Scholar 检索与被审论文最相关的近期文献:
- 使用 LLM 从论文标题和摘要中提取 3-5 组搜索关键词
- 检索最近 3 年(可配置)的相关论文
- 去重并筛选出 5-10 篇最相关的
- 标记哪些已被本文引用、哪些未被引用
- 检索结果将注入到所有审稿人的 prompt 中

如果检索失败或使用 `--no-literature` 参数,将跳过此步骤继续审稿。

### 第一步:解析论文

先运行:

```bash
uv run hypo-research review --parse-only <paper_path>
```

检查 `PaperStructure` 中的 title、abstract、sections、figures、tables、claims、references、inferred_domain 和 raw_text。若解析明显异常,先向用户说明并确认输入文件。

### 第二步:确认审稿配置

确认目标 venue、审稿团、苛刻程度和论文领域。用户已明确给出参数时,可以直接继续。

### 第三步:逐角色审稿

生成角色 prompt:

```bash
uv run hypo-research review <paper_path> \
  --venue <venue> \
  --panel <panel> \
  --severity <severity> \
  --prompts
```

依次切换为每个审稿人角色,基于 prompt 进行审稿。每个角色独立审稿,输出结构化审稿意见。

**重要**:审稿时你要完全进入该角色的性格和视角。每个 Weakness 必须同时给出具体问题、改进建议和重要程度理由。

审稿人角色:
- 🏛️ 贺云翔:Senior AC,关注 novelty、算法深度和本质贡献。
- 🔬 李超凡:Expert-1,关注技术正确性、实验公平性、SOTA 对比和 claim 是否 over。
- 🔬 吴浩宇:Expert-2,关注相邻领域盲点、跨领域可行性和遗漏相关工作。
- 📐 陈泉宇:Related,关注可读性、术语解释、figure 是否自解释。
- 🤔 蒋烨:Outsider,关注故事线、motivation、contribution 和整体自洽性。
- ✍️ 李宇轩:Writing,关注表达、格式、术语一致性和引用规范;不打数值分。
- 🔧 丁麒涵:Experimental Rigor,关注实验设计、baseline/ablation、公平性、指标定义、workload、仿真/测量设置和 claim 支撑力度;硬件论文不因未公开 RTL/完整代码本身扣分。

### 第三点五步:AC Meta-Review(如果贺云翔在 panel 中)

在所有审稿人完成独立审稿后,切换回贺云翔 AC 角色:
- 综合所有审稿意见
- 找出共识和分歧
- 给出最终建议和修改优先级
- 输出 `MetaReview` 结构

### 第三点六步:修改路线图(如果贺云翔在 panel 中)

在 Meta-Review 完成后,切换为“导师”角色:
- 站在作者一边,将审稿意见分类为必改、建议改、可忽略
- 为每个必改项给出具体方案和工作量预估
- 生成修改时间表和问题交叉矩阵
- 输出 `RevisionRoadmap` 结构

### 第四步:一致性检查(始终执行)

对所有审稿意见进行规则检查:
- 检查是否有泛泛而谈的意见(未引用具体章节、图表或公式)
- 检查是否有自相矛盾的意见
- 生成 `ConsistencyReport`,附在报告末尾

### 第五步:汇总报告

将所有 `SingleReview`、`MetaReview`、`RevisionRoadmap` 和 `ConsistencyReport` 汇总为最终 Markdown 报告。CLI 可生成报告框架:

```bash
uv run hypo-research review <paper_path> --output review_report.md
```

机器可读输出:

```bash
uv run hypo-research review <paper_path> --json
```

## 输出格式

每个审稿人输出:
1. Summary(2-3 句)
2. Strengths(编号列表)
3. Weaknesses(编号列表,标注 `[Major]` 或 `[Minor]`,且必须可操作)
4. Questions to Authors(编号列表)
5. Missing References(如有)
6. Score: X/10(Writing 角色不打分)
7. Decision: Strong Accept / Accept / Weak Accept / Borderline / Weak Reject / Reject / Strong Reject
8. Confidence: X/5

最终报告必须包含总体评分表、AC Meta-Review(如适用)、Major/Minor/Strengths 汇总、详细审稿、修改路线图(如适用)和一致性检查。

## 常用命令

From HypoxanthineOvO/Hypo-Research, MIT.

More from HypoxanthineOvO/Hypo-Research

Signals

Stars
2
Forks
0
Open issues
0
Last commit (this folder)
May 9, 2026 (5 months ago)
Last release
May 9, 2026 (5 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from HypoxanthineOvO

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get hypo-review running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying hypo-review?

CallGet help
Cite this page

hypo-review skill. Agent Mart. https://agentmart.dev/skills/hypoxanthineovo-hypo-review (last updated 2026-10-24)