← Back to Skills Marketplace
46
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install ai-research-tools
Description
整合多智能体协作与接口标准化，实现科研数据快速获取、论文复现、审稿回复及报告自动生成，提升AI辅助学术效率。
README (SKILL.md)
AI 科研工具全景指南
论文复现 · 自主研究 · 审稿提效 · 多智能体协作
一、核心定位
本技能整合 AI 科研工具完整生态，覆盖：
- 数据获取提速 · 论文复现实战 · 审稿回复提效 · 多智能体系统 · 聚合平台
⚡ 核心理念：AI 是科研效率放大器，承担耗时基础工作，研究者保留核心判断。
二、数据获取提速
2.1 MCP 接口方案
| 项目 | 详情 |
|---|---|
| 核心能力 | 对接国内外多类学术、政府数据源 |
| 效率提升 | 找数据：1个月 → 1次 MCP 操作 |
| 代表工具 | MCP 协议（Model Context Protocol） |
MCP 优势：
- 标准化接口，一次配置多源调用
- 支持中文环境（人大 DeepAnalyze）
- 实时更新数据源
2.2 人大 DeepAnalyze
| 项目 | 详情 |
|---|---|
| 出品 | 中国人民大学 |
| 核心能力 | 自主完成：找数据 → 跑分析 → 写报告 |
| 适用环境 | 中文环境（国内数据/政策研究） |
典型流程：
用户输入研究问题 → DeepAnalyze 自动找数据
→ 跑统计分析 → 生成完整报告 → 输出结论
三、论文复现实战
3.1 Stata Skill 复现方案
| 项目 | 详情 |
|---|---|
| 速度 | 10 分钟复现 AER 顶刊论文 |
| 方法 | Stata 代码 → Python 转译 |
| 验证结果 | 核心系数、标准误、显著性水平全部匹配 ✅ |
技术细节：
- 保留原始 Stata 逻辑
- Python 语法等价转换
- 自动对标显著性阈值
3.2 三大自主研究系统
🧪 Agent Laboratory
| 项目 | 详情 |
|---|---|
| Star | 5500+ ⭐ |
| 效果 | 降低研究成本 84% |
| 定位 | 端到端研究助手 |
🧪 AI-Scientist-v2
| 项目 | 详情 |
|---|---|
| 突破 | 首篇 AI 撰写且通过同行评审的论文 ✅ |
| 意义 | AI 科研产出获学术界认可 |
🧪 ARIS
| 项目 | 详情 |
|---|---|
| 能力 | 隔夜完成自主研究 |
| 适用 | 紧急研究任务 / 快速文献综述 |
四、审稿回复提效
4.1 三大提效工具
📝 review-response
| 功能 | 说明 |
|---|---|
| 自动分类 | 智能识别审稿意见类型 |
| 起草回复 | 按类别生成回复草稿 |
| 效率提升 | 审稿周期 6-8 个月 → 2 周 |
📝 ai-research-feedback
| 功能 | 说明 |
|---|---|
| AI 审稿模拟 | 6 个 AI 代理模拟顶刊审稿流程 |
| 覆盖维度 | 方法论 / 数据 / 创新性 / 写作 |
📝 paper-slide-deck
| 功能 | 说明 |
|---|---|
| 一键转 PPT | 论文内容自动生成演示文稿 |
| 适用场景 | 会议报告 / 答辩 / 组会分享 |
4.2 审稿回复周期对比
传统方式：6-8 个月
↓
AI 辅助：2 周
效率提升：10-20x
五、多智能体系统
5.1 角色分离分工模式
CoPaper.AI
Supervisor（统筹）
├── 选题代理
├── 分析代理
├── 写作代理
└── 审核代理
优势：各司其职，减少 AI "幻觉"，提高整体质量
港大 AI-Researcher
| 能力 | 说明 |
|---|---|
| 端到端 | 从文献 → 论文全流程 |
| 适用 | 英文论文 / 国际期刊 |
人大 DeepAnalyze
| 能力 | 说明 |
|---|---|
| 中文环境 | 国内数据/政策研究 |
| 自主实证 | 自动建模 + 分析 + 报告 |
5.2 多智能体架构优势
| 传统方式 | 多智能体协作 |
|---|---|
| 单 AI 输出，质量不稳定 | 角色分工，互相校验 |
| 人工协调多个工具 | Supervisor 自动调度 |
| 容易出现 AI 幻觉 | 审核代理降低错误率 |
六、聚合生态平台
6.1 三大平台对比
| 平台 | Skills 数量 | Star 数 | 特色 |
|---|---|---|---|
| ClawHub | 13,729+ | - | OpenClaw 官方生态 |
| VoltAgent | 5,400+ | 44,791 ⭐ | 多智能体框架 |
| antigravity | 1,340+ | 30,578 ⭐ | 轻量级聚合 |
6.2 配套资源
| 资源 | 说明 |
|---|---|
| 官方指南 | 快速上手教程 |
| 学术讲座 | 视频教学 |
| 因果推断入门教材 | 方法论基础 |
七、AI vs 人类分工
7.1 AI 承担的工作
| 类型 | 占比 | AI 能力 |
|---|---|---|
| 数据查找 | 80% | MCP 接口自动抓取 |
| 代码转译 | 90% | Stata → Python |
| 文献整理 | 70% | 自动摘要 + 分类 |
| 统计分析 | 75% | 自动化建模 |
| 审稿分类 | 80% | NLP 智能分类 |
| PPT 生成 | 85% | 一键转换 |
7.2 人类保留的工作
| 工作 | 原因 |
|---|---|
| 研究问题定义 | 需要领域直觉和创造力 |
| 结论解读 | 需要判断实际意义 |
| 学术贡献定性 | 需要主观评估创新性 |
| 同行评审 | 人类信用背书 |
八、使用场景速查
| 场景 | 推荐工具 |
|---|---|
| 复现 AER 顶刊论文 | Stata Skill + Python 转译 |
| 快速文献综述 | ARIS（隔夜完成） |
| 审稿回复 | review-response + ai-research-feedback |
| 论文转 PPT | paper-slide-deck |
| 中文实证研究 | 人大 DeepAnalyze |
| 全流程自主研究 | Agent Laboratory |
| AI 生成论文（首篇同行评审） | AI-Scientist-v2 |
| 多智能体协作写论文 | CoPaper.AI |
| 数据获取 | MCP 接口 |
九、快速上手指南
第一步：配置数据源（MCP）
# 安装 MCP 工具
npx clawhub@latest install mcp-tools
# 配置学术数据源
openclaw mcp add academic --connector arxiv --connector semantic-scholar
第二步：选择复现工具
# 安装 Stata Skill
npx clawhub@latest install stata-skill
# 使用 Python 转译
stata-to-python --input paper.do --output paper.py
第三步：使用审稿工具
# 安装审稿回复助手
npx clawhub@latest install review-response
# 一键转 PPT
npx clawhub@latest install paper-slide-deck
十、注意事项
✅ 正确姿势：
- AI 承担耗时基础工作
- 人类保留核心判断
- 交叉验证 AI 生成的结论
⚠️ 注意事项：
- AI 可能产生"幻觉"，需人工审核
- 论文中使用 AI 生成内容需声明
- 遵守各平台使用条款
十一、与其他技能的关联
| 本技能 | 关联技能 | 关系 |
|---|---|---|
| AI科研工具 | mckinsey-frameworks |
战略分析框架辅助研究设计 |
| AI科研工具 | thinking-knowledge-system |
思考四层次（问题定义） |
| AI科研工具 | knowledge-system-guide |
知识体系构建（文献管理） |
| AI科研工具 | investor-reading-list |
投资研究（金融论文复现） |
十二、使用方式
触发场景
用户说「复现一篇 AER 论文」→ Stata Skill + 转译流程
用户说「快速做文献综述」→ ARIS 工具
用户说「帮我回复审稿意见」→ review-response
用户说「论文转 PPT」→ paper-slide-deck
用户说「AI 能帮我做研究吗」→ 展示完整工具链
用户说「多智能体研究系统」→ CoPaper.AI / 港大 AI-Researcher
组合使用
用户：「我想做一篇关于中国资本市场的实证论文」
→ 人大 DeepAnalyze（数据 + 分析）
→ Stata Skill（复现方法）
→ review-response（审稿回复）
→ paper-slide-deck（答辩展示）
本技能整合 AI 科研工具完整生态，助力高效学术研究
Usage Guidance
This SKILL.md is a high‑level guide (no bundled code), but before running any of its recommended commands you should: 1) recognize the guide expects command-line tools (node/npx, a CLAWhub/openclaw CLI, and other tool binaries) even though they are not declared — ensure you have or trust these tools; 2) verify the provenance of packages you will fetch with npx (publisher, npm page, GitHub repo, release signatures) — npx will download and execute code on your machine; 3) be prepared to supply connector API keys or credentials when configuring data sources and avoid pasting sensitive secrets unless you trust the connector and understand its privacy policy; 4) prefer installing packages in a sandbox or VM if you are unsure; and 5) double-check that institutional or legal constraints allow automated scraping/aggregation of the data sources you plan to use. If you want, I can list the exact commands the guide uses and suggest safer alternatives or checks to validate the packages and CLIs before installing.
Capability Analysis
Type: OpenClaw Skill
Name: ai-research-tools
Version: 1.0.0
The skill bundle is a comprehensive informational guide for using AI tools in academic research, covering data acquisition, paper replication, and review management. The shell commands provided in SKILL.md (e.g., npx clawhub@latest install) are standard installation procedures for the OpenClaw ecosystem, and the instructions are clearly aligned with the stated purpose of assisting researchers without any evidence of malicious intent, data exfiltration, or prompt injection.
Capability Assessment
Purpose & Capability
The SKILL.md is an instruction-only guide describing how to compose and use many research tools and installers (npx, openclaw, stata-to-python, etc.). The skill metadata declares no required binaries or env vars, but the runtime instructions assume the presence of command-line tools (node/npx, openclaw CLI, stata-to-python) and ability to install packages. This gap (undeclared required tooling) is an incoherence the user should notice.
Instruction Scope
Instructions stay within the research-tooling domain (install connectors, translate Stata -> Python, run review helpers). They do instruct the agent/user to install and invoke third-party tools and to operate on user files (e.g., paper.do), which is expected for this purpose but grants the guide broad discretion to run downloaded code. The guide does not instruct reading unrelated system files or secrets, but connector setup may later request API keys (not documented here).
Install Mechanism
There is no install spec in the skill package itself (lowest static risk), but SKILL.md instructs use of npx to fetch and install packages (e.g., `npx clawhub@latest install ...`) and to run `openclaw mcp add`. Using npx/openclaw will download and execute third‑party code at runtime — normal for an installer guide but higher risk than pure read‑only guidance. The skill does not identify package origins or verify signatures.
Credentials
The skill declares no required environment variables or credentials, which is consistent with being a high-level guide. However, many referenced connectors and tools (MCP connectors, academic APIs, platform CLIs) commonly require API keys or tokens during setup; the guide does not mention these or how they should be handled. This omission reduces transparency about what secrets you may need to provide later.
Persistence & Privilege
The skill is not always-enabled, does not request elevated or persistent agent privileges, and contains no code writing to disk as part of the packaged skill. The only persistence/side effects come from the external installers the guide instructs you to run.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install ai-research-tools - After installation, invoke the skill by name or use
/ai-research-tools - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: Comprehensive AI research tools panorama guide for efficient academic workflows.
- Covers major areas: rapid data access, paper reproduction, peer review assistance, multi-agent collaboration, and ecosystem platforms.
- Details leading tools and protocols such as MCP, DeepAnalyze, Stata Skill (with Stata-to-Python translation), Agent Laboratory, and review-response.
- Includes best practices, human-vs-AI workflow division, ecosystem comparisons, and quickstart instructions.
- Lists use-case scenarios and integration with related skills for streamlined research processes.
Metadata
Frequently Asked Questions
What is AI科研工具全景指南?
整合多智能体协作与接口标准化，实现科研数据快速获取、论文复现、审稿回复及报告自动生成，提升AI辅助学术效率。 It is an AI Agent Skill for Claude Code / OpenClaw, with 46 downloads so far.
How do I install AI科研工具全景指南?
Run "/install ai-research-tools" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is AI科研工具全景指南 free?
Yes, AI科研工具全景指南 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does AI科研工具全景指南 support?
AI科研工具全景指南 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created AI科研工具全景指南?
It is built and maintained by SMS (@smseow001); the current version is v1.0.0.
More Skills