# arxiv-daily

> Fetch and filter top papers from arXiv based on configurable topics, send to Feishu

- Author: randy
- Repository: skywalker8210/ArxivDaily
- Version: 20260130214205
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-06
- Source: https://github.com/skywalker8210/ArxivDaily
- Web: https://mule.run/skillshub/@@skywalker8210/ArxivDaily~arxiv-daily:20260130214205

---

---
name: arxiv-daily
description: Fetch and filter top papers from arXiv based on configurable topics, send to Feishu
user-invocable: true
---

# ArxivDaily Skill

This skill automatically fetches the latest papers from arXiv based on configured topics and sends formatted notifications to Feishu.

## Usage

### Manual Invocation

Users can invoke this skill via slash command in Feishu:

```
/arxiv-daily [topic] [--count N] [--days D] [--lang zh|en]
```

**Parameters:**
- `topic` (optional): Specific topic keyword to search for. If omitted, uses all configured topics.
- `--count N` (optional): Number of papers to fetch. Default: 5
- `--days D` (optional): Number of days to look back. Default: 1
- `--lang zh|en` (optional): Language for the message. Default: zh (Chinese)

**Examples:**
- `/arxiv-daily` - Get 5 papers from all configured topics (last 1 day)
- `/arxiv-daily KV Cache --days 3` - Get papers about KV Cache from the last 3 days
- `/arxiv-daily --count 10 --days 7 --lang en` - Get 10 papers from the last week in English
- `/arxiv-daily inference optimization --count 3` - Get 3 papers about inference optimization

### Automatic Execution

This skill can also be triggered automatically via cron job for daily scheduled pushes.

## How It Works

1. **Fetch Papers**: Queries arXiv API for recent papers in configured categories (cs.AI, cs.LG, cs.CL, etc.)
2. **Filter & Rank**: Calculates relevance scores based on keyword matching in titles and abstracts
3. **Deduplicate**: Checks cache to avoid sending duplicate papers
4. **Format Message**: Generates rich text Feishu message cards with paper details
5. **Send to Feishu**: Uses OpenClaw's Feishu integration to deliver the message

## Configuration

Edit `{baseDir}/config/topics.yaml` to customize:
- Topics and keywords to track
- arXiv categories to search
- Minimum relevance score threshold
- Default language and paper count
- Schedule time

## Scripts

- `{baseDir}/scripts/fetch_papers.py` - arXiv API integration
- `{baseDir}/scripts/filter_papers.py` - Paper filtering and ranking
- `{baseDir}/scripts/format_message.py` - Feishu message formatting
- `{baseDir}/scripts/main.py` - Main entry point

## Requirements

- Python 3.8+
- Dependencies: requests, pyyaml, python-dateutil

Install with:
```bash
cd {baseDir}
pip3 install -r requirements.txt
```