agentic-paper-digest-skill

What problem does it solve?

It helps users quickly discover and summarize new arXiv and Hugging Face research by turning continuous paper streams into a structured, machine-readable digest.

Core Features & Use Cases

• Automated Paper Fetching & Recency Windows: Pulls recent papers over a configurable time window (default 24 hours) and filters results to keep only relevant items.
• LLM-Based Relevance + Summarization: Uses separate LLM calls/models for relevance filtering and for generating summaries mapped to predefined topic IDs.
• JSON and API Workflows: Produces JSON output for downstream automation or runs an optional local API server to poll for results and query stored papers.
• Configurable Topics and Caps: Uses config files to define topic IDs, descriptions, keywords, and per-topic result limits, plus controls for source paging and candidate limits.

Quick Start

Ask the AI to fetch a 24-hour digest from arXiv and Hugging Face and return the results as JSON using the locally bootstrapped CLI workflow.