How arnnv/email-tracking-system-langgraph works
Project Type & Purpose
• The project is a full-fledged email processing and tracking system.
• It automatically downloads emails, categorizes them (e.g., spam, job, urgent, general), uses agents to perform summarization and classification, extracts job details, sends notifications, and provides a Streamlit dashboard for interactive management.
• The README clarifies that it employs LangGraph workflows and has an agent-based modular design.
File Structure
• Key directories:
• agents/ – Contains specialized agents (e.g., classification_agent and summarization_agent) that encapsulate distinct responsibilities.
• core/ – Provides the essential services like email downloading (email_downloader.py), fetching (email_fetcher.py), database initialization, and utilities.
• processing/ – Contains modules for handling different categories of emails (general, job, spam, urgent) and the module to fetch emails.
• models/ – Data models defining email structures and state.
• workflows/ – Defines processing workflows with LangGraph, with modules like graph_builder.py.
• jobs/ – Manages job application details with applications, tracking, and viewing capabilities.
• prompts/ – Houses language model prompt templates used for classification, summarization, and job detail extraction.
• Top-level scripts:
• app.py – Launches the Streamlit web dashboard.
• main.py – CLI entry point.
• apply_for_job.py and view_applications.py – Additional job application handling scripts.
• Other important files include configuration (config.py), dependency list (requirements.txt), and deployment/configuration files in the .streamlit folder.
Elements
• External integrations (IMAP server, Ollama for LLMs).