Claude Powered Study Assistant

July 18, 2024 · View on GitHub

This is a Python-based study assistant powered by an LLM (Large Language Model). It provides various tools to assist with different tasks related to research, coding, note-taking, and more. Below is a description of each tool along with its capabilities and usage instructions.

chart

Tools Available

  1. Wikipedia Search
    • Description: Searches Wikipedia based on user input. Useful for scientific or specific inquiries.
  2. DuckDuckGo Search
    • Description: Utilizes DuckDuckGo's Search API to find information on the internet. Useful for general inquiries.
  3. Note Saving
    • Description: Saves notes to 'notes.txt' for further user access.
  4. Code Executer
    • Description: Executes the provided code string and compares the result to the expected output.
  5. Calculator
    • Description: Basic arithmetic calculator supporting addition, subtraction, multiplication, and division.
  6. YouTube Search
    • Description: Searches YouTube based on the given input and returns the most related 5 videos.
  7. File Reading
    • Description: Extracts text from various file formats (PDF, TXT, DOCX).

Usage

  1. Import the Tool class from Tool.py.
  2. Create instances of each tool with appropriate parameters.
  3. Use the tools as needed by passing required parameters to their respective functions.

Example:

from Tool import Tool

# Create tool instances
wiki = Tool(tool1_name, tool1_description, tool1_parameters)
search = Tool(tool2_name, tool2_description, tool2_parameters)
# Create more instances for other tools...
TOOLS = [wiki, search, ..]

# Use the tools
 names = [tool.name for tool in TOOLS]
 descriptions = [tool.description for tool in TOOLS]
 parameters = [tool.parameters for tool in TOOLS]
 all_tools = construct_format_tool_for_claude_prompt(names, descriptions, parameters)
 system_prompt = construct_tool_use_system_prompt([all_tools]))

function_calling_message = client.messages.create(
        model=MODEL_NAME,
        max_tokens=1024,
        messages=[message],
        system=system_prompt,
        stop_sequences=["\nHuman:", "\nAssistant", "</function_calls>"]
    ).content[0].text + '</function_calls>'
print(function_calling_message)

Example prompts and outputs:

Prompt = "can you read 'Cihan Yalçın.pdf' and summarize it for me."

pdf

Prompt = "search YouTube for Machine Learning"

pdf