How raoulbia-ai/langraph-research-assistant-blogger works
Project Type and Purpose
• This is a command-line “Research Assistant Blogger” application (MVP) that:
• Lets a user search ArXiv for papers by topic
• Uses LLMs (via OpenAI API) to analyze paper content
• Generates a technical blog post
• Builds and visualizes a paper relationship graph (LangGraph)
• User Interface (CLI in app.py)
• Core Logic (src/main.py)
• Graph Module (src/graph and graph/)
• External Data Clients (utils/arxiv_client.py, utils/google_scholar_client.py)
• LLM Service (OpenAI API)
• Configuration (.env or config.json)
• Storage (local file system for saving posts)
• Tests (tests/)
Relationships and Data Flow
• User → CLI (app.py)
⇨ invokes Core Logic
⇨ Core Logic calls Graph Builder and Workflow nodes
⇨ Workflow nodes call External Data Clients to fetch papers
⇨ Graph Module organizes papers into nodes/edges
⇨ Core Logic calls OpenAI API for analysis/generation
⇨ Generated blog post is written to disk
• Config values (API keys, endpoints) flow into clients and LLM calls
Architectural Patterns and Technologies
• Modular Python packages (src/, graph/, utils/)
• Workflow/Node pattern via LangGraph (graph/workflow.py, nodes.py)
• External service integration (ArXiv API, Google Scholar, OpenAI)
• CLI entry point (app.py)