Planning Agent

简体中文

Project Overview

Planning-Agent is an intelligent learning assistant system built with FastAPI. It provides personalized learning planning based on user profiles. Core features include:

• Collecting basic user information (user_portrait) (not implemented yet)
• Receiving user learning prompts (prompt)
• Generating personalized study paths, including:
  • Learning goals (learning_goal)
  • Detailed learning plans (learning_plan)
  • Custom test questions (exam_questions)

Agent Graph

---
config:
  flowchart:
    curve: linear
---
graph TD;
        __start__([<p>__start__</p>]):::first
        input(input)
        planner(planner)
        evaluator(evaluator)
        examiner(examiner)
        __end__([<p>__end__</p>]):::last
        __start__ --> input;
        evaluator --> planner;
        input --> planner;
        planner -.-> evaluator;
        planner -.-> examiner;
        examiner --> __end__;
        classDef default fill:#f2f0ff,line-height:1.2
        classDef first fill-opacity:0
        classDef last fill:#bfb6fc

Loading

Environment Setup

Backend

To clone the entire repository including submodules, run:

git clone --recursive https://github.com/Blattvorhang/Planning-Agent.git

If you've already cloned the repo without --recursive, initialize the submodules manually:

git submodule update --init --recursive

Install the required Python packages:

pip install -r requirements.txt
uv pip install -e "./graph/tools/arxiv-mcp-server/[test]"

Then, configure your LLM API key in the .env file. Use .env.example as a reference.

Frontend

Install dependencies by running:

cd frontend
npm install

Running the Project

Currently, the project supports local deployment only. Backend runs at localhost:8000, and the frontend is available at localhost:3000.

Start the Backend

You can launch the FastAPI backend using either of the following commands:

uvicorn main:app --reload
# or
python main.py

Start the Frontend

Run:

cd frontend && npm run dev

API Reference

Learning Plan Endpoint

• Endpoint: POST /api/learn

• Request Body:

  {
    "prompt": "Description of the learning goal"
  }

• Response:

  {
    "learning_goal": "Generated learning goal",
    "answer": "Learning plan",
    "exam_questions": "Test questions"
  }

Project Structure

Planning-Agent/
├── LICENSE
├── README.md
├── README_zh.md
├── app
│   ├── README.md
│   ├── __init__.py
│   ├── api.py         # FastAPI route definitions
│   ├── deps.py        # Dependency injection
│   ├── models.py      # Data models
│   ├── question.py    # Data structure for questions
│   └── service.py     # Core service logic
├── assets
│   └── demo.mp4
├── chatbot
├── frontend
├── graph
│   ├── agents
│   ├── tools          # Tool integrations
│   └── workflow
├── main.py            # Entry point for the FastAPI app
└── requirements.txt   # Dependency requirements

Extended Features

This project integrates with the arxiv-mcp-server tool, enabling:

• Academic paper search
• Paper content downloads
• Research material analysis

For detailed usage, refer to tools/arxiv-mcp-server/README.md.