arXiv Daily Paper Pusher#

Description#

Automatically fetch yesterday's arXiv papers, rank by keyword relevance, and push to Feishu (Lark) via webhook.

Features#

• Multi-group support with independent keywords
• Smart scoring: title matches weighted 2x, abstract 1x
• Dual-mode API: arxiv library with automatic HTTP fallback
• Per-group or single message push strategies

Schedule#

yaml

schedule: "30 2 * * *"

Runs daily at 10:30 AM Beijing Time (02:30 UTC).

Requirements#

• Python 3.10+
• Dependencies: arxiv, PyYAML, requests
• Feishu Incoming Webhook URL

Quick Start#

1. Install:

bash

pip install -r requirements.txt

1. Configure:

bash

cp config.example.yaml config.yaml
# Edit config.yaml with your webhook URL and keywords

1. Run:

bash

python main.py

Configuration#

See config.example.yaml for full configuration options:

• groups: Research groups with keywords
• feishu_webhook: Your Feishu bot webhook URL
• top_k: Papers per group (default: 6)
• timezone_offset: Hours from UTC (default: 8 for Beijing)
• api_mode: "auto" | "arxiv_only" | "http_only"
• push_strategy: "per_group" | "single"

Manual Test#

bash

cd ~/.openclaw/skills/arxiv-daily-pusher
python main.py