Skip to content

README.md

Latest commit

 

History

History
212 lines (151 loc) · 6.83 KB

README.md

File metadata and controls

212 lines (151 loc) · 6.83 KB

🎓 Learning Assistant

An intelligent conversational learning companion that integrates with your Obsidian knowledge base to provide contextual responses and automatically generate session summaries.

Tool-Based Workflow

The assistant now uses a tool-based workflow, where the LLM suggests tool calls and the Python backend executes them. This enables:

  • Modular tool functions for context search, session saving, and more
  • Clear separation between LLM reasoning and backend actions
  • Easier extensibility for new tools and features

How it works:

  • The LLM returns a tool call (e.g., chat_with_context_tool or save_session_tool)
  • The backend inspects the tool call, executes the corresponding Python function, and returns results
  • The LLM then uses the tool results to generate a final conversational response

Configuration Options

You can configure the assistant using the .env file and Python settings:

  • ANTHROPIC_API_KEY: Claude API key
  • VOYAGE_API_KEY: Voyage AI embeddings key
  • OBSIDIAN_VAULT_PATH: Path to your Obsidian vault
  • LLM_MODEL: Claude model (e.g., claude-3-haiku-20240307)
  • LLM_TEMPERATURE: LLM response creativity (float, e.g., 0.2)
  • EMBEDDING_MODEL: Voyage model (e.g., voyage-3-lite)
  • TOP_K: Number of relevant documents to retrieve (default: 3)
  • CHUNK_SIZE: Document chunk size for indexing (default: 512)
  • CHUNK_OVERLAP: Overlap between chunks (default: 50)

These can be set in .env or in the relevant Python config files.

Updated Project Structure

The project is organized for clarity and modularity:

LearningAssistant/
├── main.py                     # Application entry point & tool orchestration
├── requirements.txt            # Python dependencies
├── .env                        # Environment variables
├── services/
│   ├── llm_service.py          # LLM integration & prompt management
│   ├── vector_store.py         # ChromaDB & Voyage AI embeddings
│   ├── obsidian_service.py     # Obsidian vault operations
├── agent/
│   ├── learning_agent.py       # Agent logic
│   └── tools/
│       ├── chat.py             # Context search tool
│       ├── storage.py          # Session save tool
│       └── analysis.py         # Analysis tool helper
├── core/
│   ├── config.py               # Configuration management
│   ├── conversation.py         # Conversation state/history
├── utils/
│   └── prompt_templates.py     # LLM prompt templates
├── tests/
│   ├── test_integration_basic.py
│   └── test_integration_complete.py
└── chroma_db/                  # Vector database storage

Session Summary Format

Session summaries are automatically saved to your Obsidian vault in a daily folder structure, with unique filenames and backlinks to referenced files.

File Structure:

<OBSIDIAN_DAILY_NOTES_FOLDER>/
    July 25, 2025/
        What_is_machine_143022.md

Session Summary Example:

---
created: 2025-07-11T14:30:22
type: learning_session_summary
daily_folder: July 11, 2025
tags: [learning, session, summary]
---

# Learning Session Summary

<session summary content>

## Referenced Files

- [[Some_Note]]
- [[Another_Note]]

Backlinks:

  • For every referenced file, a backlink is automatically added to the original note under a ## References section:

    ## References

- [[Daily Notes/July 11, 2025/What_is_machine_143022]]

File Naming:

  • Session summaries are named as <SessionName>_<HHMMSS>.md for uniqueness and clarity.

This structure ensures easy navigation, traceability, and rich interlinking between your learning sessions and your existing notes.

✨ Features

  • 🤖 Intelligent Conversations: Chat naturally about any topic with Claude AI
  • 🔍 Knowledge Base Integration: Searches your Obsidian vault for relevant context
  • 📝 Automatic Note Generation: Creates comprehensive session summaries
  • 🔄 Seamless Integration: Saves learning sessions back to your Obsidian vault
  • 💡 Context-Aware Responses: Maintains conversation flow and references your existing knowledge
  • ⚡ Real-time Search: Vector-based search through your personal knowledge base

Core Components

  • Vector Store: ChromaDB with Voyage AI embeddings for semantic search
  • LLM Service: Claude Sonnet 4 for advanced, intelligent responses
  • Obsidian Integration: Direct file system integration with markdown generation

🖥️ GUI Functionality

The Learning Assistant now includes a modern, chat-like graphical user interface (GUI) built with PySide6. This GUI provides a seamless and interactive experience for managing conversations, viewing context, and saving notes.

Key Features

  • Chat Display: Messages are shown in a visually appealing, chat-style format with Markdown rendering (including code blocks).
  • Message Alignment: User and assistant messages are aligned for clarity, with consistent font sizing and spacing.
  • Scroll & Focus: The chat window automatically scrolls to the latest message and keeps the input field focused for fast interaction.
  • Session State: The GUI maintains session state and displays confirmation messages when notes are saved.
  • Backend Integration: The backend runs in a separate process for stability and responsiveness.
  • Tool Calls: Tool calls and results are handled transparently, with only relevant conversational output shown to the user.

How to Use

  1. Start the GUI by running: 
    python gui.py
  2. Interact with the assistant in the chat window. All features available in the CLI are supported, plus:
    • Rich Markdown rendering
    • Automatic scroll-to-bottom
    • Confirmation and introductory messages
    • Note saving and session management

The GUI is designed to closely resemble modern messaging apps, making your learning experience more intuitive and enjoyable.

🚀 Quick Start

1. Prerequisites

  • Python 3.8+
  • Obsidian vault with markdown files
  • API keys for Anthropic (Claude) and Voyage AI

2. Installation

# Clone the repository
git clone <repository-url>
cd LearningAssistant

# Install dependencies
pip install -r requirements.txt

3. Environment Configuration

Create a .env file in the root directory:

ANTHROPIC_API_KEY=your_anthropic_key_here
VOYAGE_API_KEY=your_voyage_key_here
OBSIDIAN_VAULT_PATH=C:\path\to\your\obsidian\vault

4. Build Vector Index

python services/build_index.py

5. Start the Assistant with GUI

python gui.py

🤝 Contributing

  1. Fork the repository
  2. Create a feature branch
  3. Add tests for new functionality
  4. Ensure all tests pass
  5. Submit a pull request

Ready to enhance your learning journey? Start exploring with the Learning Assistant today! 🚀