According to AlphaMoat, as of 2026-09, arXiv 论文推荐器 has 1,906 downloads and 1 stars — ranked #4,489 of 5,078 AI skills overall, and #394 of 465 in Professional.
| Platform | Installs | Downloads | Stars | Comments |
|---|---|---|---|---|
| ClawHub ↗ | 25 | 993 | - | - |
| SkillHub ↗ | 282 | 1.9K | 1 | — |
Platforms count differently and update at different times.
Recommends recent Agent/RAG papers with companion code and aims to help readers grasp their main contribution in one minute. Four topics are supported: agent-eval, rag-eval, agent-arch, rag-arch — random if unspecified. Workflow: recommend.py [topic] will 1. searches arXiv papers from the last 6 months, visit paper pages to extract GitHub links, verify repository validity, analyze abstracts, and save a Markdown report to ~/papers/recommendations/; recommendations include title, topic, date, GitHub stars, a one-line takeaway, and reasons. Quality control: - ✅ requires a GitHub code link and a publicly available repository; a memory file (~/papers/history.json) avoids repeat recommendations. Requires Python 3.8+ and the free arXiv/GitHub APIs. Card caution: fetched metadata, abstracts, and links may be inaccurate or stale — treat reports as untrusted research assistance and verify claims before acting; a version discrepancy exists (card 1.1.0 vs SKILL.md footer 1.0).
Profile published 2026-10-29
Generated from the skill’s documentation as examples; not results of actual runs.
The counting method changed in 2026-09: skills with the same name are now counted separately by author, and the full catalog is covered. Dashed lines show months before the change; they are not compared with later months, and no growth is computed across the change.
| Month | Downloads | MoM | Stars |
|---|---|---|---|
| 2026-09 | 1.9K | — | 1 |
| 2026-08 | 1.3K | +16.2% | 1 |
| 2026-07 | 1.1K | +20.1% | 1 |
| 2026-06 | 945 | +16.5% | 1 |
| 2026-05 | 811 | — | 1 |
| 2026-03 | 455 | — | - |
The author’s original file, shown as plain text without edits or review by AlphaMoat. Links are not clickable; check commands before running them. Fetched 2026-10-23.
--- name: arxiv-paper-recommender description: 推荐高质量的 Agent/RAG 论文。当用户提到"推荐论文"、"arxiv论文"、"Agent论文"、"RAG论文"、"论文推荐"、"看论文"、"找论文"、"最近有什么好论文"时触发此技能。支持主题:Agent测评、RAG测评、Agent架构、RAG架构。自动验证 GitHub 代码,生成结构化报告。 --- # arXiv 论文推荐器 推荐近期高质量 Agent/RAG 领域论文,确保有配套代码,1分钟内让读者抓住核心价值。 ## 触发词
- "推荐一篇论文"、"帮我找篇论文"
- "Agent论文"、"RAG论文"
- "论文推荐"、"arxiv推荐"
- "最近有什么好论文"
- "推荐一篇 Agent/RAG 相关的论文"
## 支持的主题
| 主题代码 | 中文名称 | 说明 |
|---------|---------|------|
| agent-eval | Agent测评 | Agent 评估方法、Benchmark |
| rag-eval | RAG测评 | RAG 评估方法、Benchmark |
| agent-arch | Agent架构 | Agent 框架设计、系统构建 |
| rag-arch | RAG架构 | RAG 系统架构、框架设计 |
用户可指定主题,不指定则随机选择。
## 工作流程
### Step 1: 确定主题
```
如果用户指定了主题 → 使用指定主题
否则 → 从四个主题中随机选择一个
```
### Step 2: 执行搜索脚本
```bash
cd ~/.openclaw/workspace/skills/arxiv-paper-recommender/scripts
python3 recommend.py [topic]
```
参数:
- `topic` 可选,值为 `agent-eval`、`rag-eval`、`agent-arch`、`rag-arch` 之一
- 不传参数则随机选择
### Step 3: 读取生成的报告
脚本会自动:
1. 搜索 arXiv 最近 6 个月的论文
2. 访问论文页面查找 GitHub 链接
3. 验证 GitHub 仓库有效性
4. 分析论文摘要提取关键信息
5. 生成 Markdown 报告
报告保存位置:`~/papers/recommendations/YYYY-MM-DD_[arxiv_id].md`
### Step 4: 向用户展示推荐结果
输出格式示例:
```
📄 推荐论文:[标题]
🎯 主题:Agent架构设计
📅 发布:2026-03-01
⭐ GitHub:1.2k stars
一句话:这篇论文提出了 [方法名],解决了 [问题]。
推荐理由:
1️⃣ 解决了 [具体问题]
2️⃣ 采用 [方法],亮点是 [创新点]
3️⃣ 实践建议:[核心建议]
4️⃣ 代码质量:⭐ X stars,最近 X 天更新
🔗 arxiv.org/abs/xxxx.xxxxx
💻 github.com/xxx/xxx
完整报告已保存至:~/papers/recommendations/xxx.md
```
### Step 5: 后续交互
推荐完成后,询问用户:
```
需要我进一步帮你:
1. 📖 深入解读这篇论文?
2. 🔍 查找相关论文?
3. 🛠️ 分析代码实现?
4. 📋 生成论文阅读笔记模板?
```
## 报告结构
生成的 Markdown 报告包含:
1. **基本信息表** - arXiv ID、发布日期、作者、链接
2. **摘要** - 论文摘要(截取前 600 字)
3. **推荐理由**:
- 解决了什么问题
- 使用了什么方法
- 工程实践建议
- 代码质量评估
4. **1分钟速览** - 快速理解论文核心
5. **快速链接** - PDF、GitHub、arXiv
## 质量控制
### 论文筛选标准
- ✅ 必须有 GitHub 代码链接
- ✅ 发表时间在最近 6 个月内
- ✅ GitHub 仓库存在且公开
- ✅ 自动分析摘要提取关键信息
### 避免
- ❌ 纯理论无代码的论文
- ❌ GitHub 仓库已删除或私有
## 记忆功能
记录用户已推荐的论文,避免重复:
```json
// 保存到 ~/papers/history.json
{
"recommended": [
{
"arxiv_id": "2403.12345",
"title": "论文标题",
"date": "2026-03-15",
"topic": "Agent架构"
}
]
}
```
每次推荐前自动检查历史记录。
## 脚本说明
| 脚本 | 功能 |
|------|------|
| `search.py` | arXiv 搜索 + GitHub 链接提取 + 仓库验证 |
| `analyze.py` | 论文摘要智能分析,提取问题/方法/结论 |
| `recommend.py` | 完整推荐流程,生成报告 |
## 依赖
- Python 3.8+
- arXiv API(免费,无需密钥)
- GitHub API(公开接口)
- 网络访问
## 示例用法
**用户**:推荐一篇 Agent 论文
**执行**:
```bash
python3 recommend.py agent-eval
```
**输出**:
```
📄 推荐论文:Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously
🎯 主题:Agent测评
📅 发布:2026-03-12
⭐ GitHub:53 stars
...
```
---
*技能版本:1.0 | 创建时间:2026-03-15*
A data-interface skill that pulls A-share quotes, financials, and sector data through the AkShare library.
Query Eastmoney financial data in natural language and get Excel plus Markdown tables back.
Help organize China-patent materials, search questions, and technical-disclosure drafts.
Financial data skill: query structured market data via the westock CLI.
A local-first market research skill that reads Yahoo Finance data for quotes and fundamentals and renders charts with technical panels.
A valuation analysis skill for A-share and Hong Kong stocks built on a value investing framework.
A script for A-share quotes, intraday volume breakdowns, and portfolio tracking.
Watch watchlist quotes and technical indicators, then push alerts ranked by severity when conditions fire.
Data month: 2026-09 · Downloads, stars and installs are aggregated monthly from public skill registries (ClawHub, SkillHub). See methodology.