Tswoen/Paper-Agent

Paper-Agent 是一个面向科研人员和学生的智能论文检索与调研工具。项目基于多智能体协作架构（LangGraph），通过自然语言处理（NLP）、自动化搜索，帮助用户高效查找学术论文、分析文献内容，并进行论文调研。Paper-Agent 支持多平台集成、关键词搜索、自动分析、论文调研，提升了学术研究的效率。适用于论文写作、学术调研、科研项目管理等多种场景，是学术调研的理想助手。

What it solves

Paper-Agent automates the tedious process of academic research and literature review. It replaces manual searching across multiple databases, individual paper screening, fragmented note-taking, and the struggle of structuring a comprehensive review paper with an integrated AI-driven workflow.

How it works

The system uses a multi-agent architecture orchestrated by LangGraph to move a research topic through several stages:

  1. Search: A SearchAgent generates a retrieval plan and fetches papers from arXiv, OpenAlex, and Semantic Scholar.
  2. Read: A ReadAgent screens abstracts for relevance; relevant papers are downloaded, converted to Markdown, chunked, and stored in a ChromaDB vector database.
  3. Analyze: An AnalyseAgent performs sub-topic and global analysis to identify research trends, consensus, and gaps.
  4. Outline: A WritingOutlineAgent creates a structured outline and maps specific evidence to sections.
  5. Write: A WritingAgent drafts the review section-by-section, retrieving additional evidence when needed and performing self-review.

Who it’s for

Academic researchers, students, and professionals who need to conduct thorough literature reviews and synthesize complex research domains into structured documents.

Highlights

  • Multi-source Retrieval: Integrated connectors for arXiv, OpenAlex, and Semantic Scholar with deduplication and scoring.
  • Progressive Reading: A tiered approach that filters by abstract before committing to full-text processing and vector indexing.
  • Evidence-Based Writing: Generates a structured outline with evidence mapping to ensure the final review is grounded in actual paper content.
  • Real-time Monitoring: A Vue 3 web interface with SSE (Server-Sent Events) to track the progress of the agentic workflow in real-time.
  • Flexible Model Configuration: Ability to assign different LLM providers and model tiers to different agents based on the task complexity.

Related

  • Project
    Xueyang-Song/paper-pilot

    Paper Pilot is a desktop (Electron + React) research assistant that crawls 8+ scholarly databases, downloads PDFs, indexes them with SQLite + FTS5 + vector search, and lets you ask a local or hosted LLM for evidence‑grounded answers. All data stays on your machine; the app provides audit‑ready citation traces and a reproducible workflow for literature review.

  • Project
    dongbeixiaohuo/writing-agent

    A collaborative AI writing system that replaces one-shot generation with a multi-agent workflow involving planning, drafting, reviewing, and humanizing to produce high-quality, authentic content.

  • Project
    Yuan1z0825/nature-skills

    A library of 22 packaged AI-agent workflows ("skills") for academic research — paper reading, Nature-style writing, figure generation, reviewer responses, literature search, and patent drafting — installable into Claude Code, Codex, Chatbox, and other agents.

  • Project
    bojieli/ai-agent-book

    AI‑Agent‑Book is a multilingual open‑source textbook on building LLM‑based AI agents, paired with 109 runnable Python experiments covering everything from prompt engineering to multi‑agent collaboration.

  • Project
    P1kaj1uu/ChattyPlay-Agent

    A multi-functional AI utility hub providing integrated access to LLMs, image generation, academic paper analysis, and various productivity tools like LaTeX and Markdown editors.