A sophisticated AI-driven pipeline that fetches, filters, and summarizes the latest research papers from arXiv and Hugging Face.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agentic-paper-digest
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install agentic-paper-digest using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Agentic Paper Digest?
Agentic Paper Digest is a specialized automation tool designed for researchers and developers within the Openclaw Skills ecosystem. It bridges the gap between massive academic repositories and actionable insights by using agentic AI to evaluate paper relevance and generate high-quality summaries. This skill is particularly valuable for those who need to monitor high-velocity research areas without manually sifting through hundreds of daily uploads. By leveraging this addition to your Openclaw Skills library, you can transform raw data from arXiv and Hugging Face into a structured, curated feed that fits perfectly into your existing developer workflows.
Agentic Paper Digest Use Cases
- Staying updated on specific arXiv categories like machine learning or computer vision daily.
- Creating a structured JSON feed of recent research for downstream AI agents or databases.
- Running a local API server to provide research data for internal dashboards or polling workflows.
- Filtering a high volume of papers based on complex topic definitions and institutional affiliation boosts.
How Agentic Paper Digest Works
- The user initiates the skill by running a bootstrap script that sets up the environment and local repository.
- The system loads configurations including research topics, search limits, and API credentials for OpenAI or LiteLLM.
- A fetcher module queries arXiv and Hugging Face APIs for papers published within a specific hourly window.
- An AI agent analyzes each paper's abstract and metadata against the defined topic descriptions to determine relevance.
- If enabled, the skill extracts text from the first page of the paper's PDF to create more detailed summaries.
- Results are stored in a local SQLite database and outputted as JSON for immediate use or served via a REST API.
Agentic Paper Digest Setup
Ensure you have Python 3 and an API key for OpenAI or a LiteLLM-compatible provider. Use the following commands to get started with this skill:
# Bootstrap the repository
bash "{baseDir}/scripts/bootstrap.sh"
# Configure your environment in the .env file
OPENAI_API_KEY=your_api_key
WINDOW_HOURS=24
ARXIV_CATEGORIES=cs.CL,cs.AI,cs.LG
# Run the CLI to generate a paper digest
bash "{baseDir}/scripts/run_cli.sh" --json
Agentic Paper Digest Data Schema & Taxonomy
The skill manages research data and configurations using the following structure:
| Component | Format | Description |
|---|---|---|
| papers.sqlite3 | SQLite | Persistent store for paper metadata, relevance scores, and summaries. |
| topics.json | JSON | Defines the research topics, associated keywords, and maximum results per category. |
| settings.json | JSON | Manages global fetch limits, timeouts, and source-specific API parameters. |
| affiliations.json | JSON | Contains pattern-based weights to boost results from specific institutions or organizations. |
Agentic Paper Digest Advanced Features
- Multi-provider LLM support via LiteLLM for flexibility in summarization and classification models.
- Integrated PDF text extraction using PyMuPDF to improve summary accuracy when abstracts are insufficient.
- Advanced ranking logic using affiliation-based boosts to prioritize research from top-tier institutions.
- Full REST API support for remote execution, status monitoring, and dynamic configuration updates within Openclaw Skills environments.
SKILL.md
Loading
Related Openclaw Skills
Desktop Control Skill
A comprehensive desktop automation suite for Openclaw Skills enabling programmatic control over mouse, keyboard, and display operations.
NEAR Phishing Detector
A security-focused AI agent tool for identifying and reporting phishing threats within the NEAR Protocol ecosystem.
NEAR Content Creator
An automated content generation engine for the NEAR Protocol ecosystem that creates threads, market digests, and technical tutorials.
NEAR Agent Skills
A comprehensive toolkit for NEAR Protocol that provides AI agents with gas optimization strategies and deep on-chain analytics.
Origram
A programmatic photo sharing platform for bots that uses the L402 protocol for instant, automated Bitcoin Lightning payments.
ClawSwarm
A hierarchical agent swarm designed to solve extremely difficult research problems through collaborative aggregation and consensus.
METADATA
Featured*
