← Back to Skills Marketplace
83
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install arxiv-daily-pusher
Description
Fetches yesterday's arXiv papers, ranks by keyword relevance with weighted scoring, and pushes results to Feishu via webhook for multiple groups.
README (SKILL.md)
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
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
- Install:
pip install -r requirements.txt
- Configure:
cp config.example.yaml config.yaml
# Edit config.yaml with your webhook URL and keywords
- Run:
python main.py
Configuration
See config.example.yaml for full configuration options:
groups: Research groups with keywordsfeishu_webhook: Your Feishu bot webhook URLtop_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
cd ~/.openclaw/skills/arxiv-daily-pusher
python main.py
Usage Guidance
This skill appears to do exactly what it says: pull yesterday's arXiv papers, score them locally, and post text messages to a Feishu webhook you supply in config.yaml. Before installing or scheduling it, consider: (1) the Feishu webhook URL is effectively a secret that allows posting to that group — only use a webhook for a channel you control and rotate/revoke it if leaked; (2) the config file stores the webhook in plaintext on disk, so restrict file permissions where you run this; (3) the skill makes outbound HTTP requests to arXiv and Feishu, so run it in a network environment you trust; (4) pin dependency versions (requirements.txt is present) and review/keep dependencies updated; and (5) if you want stricter separation, run the script in a sandboxed account or container with limited privileges. Overall, nothing in the bundle is disproportionate or covert.
Capability Analysis
Type: OpenClaw Skill
Name: arxiv-daily-pusher
Version: 1.0.0
The 'arxiv-daily-pusher' skill is a legitimate tool designed to fetch research papers from arXiv and notify users via Feishu webhooks. The code (main.py, fetch_papers.py, push_feishu.py) follows standard practices for API interaction, including timeout handling and fallback mechanisms, and does not exhibit any signs of data exfiltration, unauthorized execution, or prompt injection.
Capability Assessment
Purpose & Capability
Name/description match the implementation: the code fetches arXiv (arxiv library or export.arxiv.org), ranks locally, and posts to a Feishu webhook supplied in config.yaml. No unrelated services or credentials are requested.
Instruction Scope
SKILL.md and the code limit actions to fetching arXiv data, scoring locally, and sending text payloads to the provided Feishu webhook. The runtime instructions and code do not read system secrets, other skills' configs, or send data to unexpected endpoints.
Install Mechanism
There is no automated download/install spec in the skill bundle; required Python packages are declared in requirements.txt and are typical for this task (arxiv, PyYAML, requests). No external arbitrary URL downloads or archive extraction are used.
Credentials
No environment variables or external credentials are requested. The Feishu webhook is provided in a local config file (config.yaml) which is appropriate for posting messages. The amount and kind of access requested are proportional to the stated functionality.
Persistence & Privilege
Skill is not forced-always and uses normal invocation. It does not modify other skills or system-wide settings. It will run scheduled jobs if the user configures a cron, which is expected for this use case.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install arxiv-daily-pusher - After installation, invoke the skill by name or use
/arxiv-daily-pusher - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
arxiv-daily-pusher 1.0.0
- Initial release.
- Automatically fetches and ranks yesterday's arXiv papers by keyword relevance.
- Supports Feishu (Lark) webhook integration for daily paper pushes.
- Multi-group support with independent keywords and scoring (title: 2x, abstract: 1x).
- Dual-mode API with automatic fallback between arxiv library and HTTP.
- Configurable scheduling and push strategies.
Metadata
Frequently Asked Questions
What is arXiv Daily Paper Pusher?
Fetches yesterday's arXiv papers, ranks by keyword relevance with weighted scoring, and pushes results to Feishu via webhook for multiple groups. It is an AI Agent Skill for Claude Code / OpenClaw, with 83 downloads so far.
How do I install arXiv Daily Paper Pusher?
Run "/install arxiv-daily-pusher" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is arXiv Daily Paper Pusher free?
Yes, arXiv Daily Paper Pusher is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does arXiv Daily Paper Pusher support?
arXiv Daily Paper Pusher is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created arXiv Daily Paper Pusher?
It is built and maintained by VentaoZzz (@ventaozzz); the current version is v1.0.0.
More Skills