Original Summary

A local multimodal research workspace that ingests PDFs into searchable text, tables, and figures, then answers questions with citations linked to source pages. The seven-paper MVP uses Docling, LangChain, Gemini, PostgreSQL, Chroma, Redis, FastAPI, and React/TypeScript, with an idempotent ingestion pipeline designed to grow the corpus safely.

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：GitHub · AI 新项目
  • 发布时间：2026/10/6 17:48:00