This repository is an automated paper aggregation and recommendation system, primarily serving the Chinese AI, RecSys, and Information Retrieval communities. It operates as a suite of Python scripts orchestrated by GitHub Actions, forming a content curation pipeline that fetches, ranks, and publishes academic papers and industry articles. The core functionality includes a daily arXiv pipeline that uses the DeepSeek LLM API for two-stage relevance scoring, a conference paper crawler that aggregates metadata from DBLP, and an industry practice article curator driven by GitHub Issues. The outputs are deployed as a static GitHub Pages site and pushed to Feishu (ι£δΉ¦) for notification. The architecture is a monolith of Python scripts rather than a web service. Key entry points are the GitHub Actions workflows that trigger data fetching, LLM-based ranking, HTML generation, and notification sending. The system relies heavily on external APIs (arXiv, DBLP, DeepSeek, Feishu) and is configured entirely through environment variables and workflow files. While actively maintained with daily runs and recent refactoring efforts (documented in .trae/specs/), the codebase exhibits significant technical drift. It runs on Python 3.8 (EOL), uses the deprecated OpenAI v0 API, and contains deprecated GitHub Actions syntax. Despite this, several components are well-crafted and demonstrate reusable patterns, such as the async crawler, the LLM ranking pipeline, and the batch job state tracker. The project is a strong example of a personal automation tool that effectively solves a specific content curation problem, but its reliance on deprecated technologies limits its long-term maintainability and reusability without significant updates.