AIBTC AI Agent Crew

Description

Caution

This is an early experiment, agents do automatic things, here be dragons, run at your own risk.

AIBTC AI Agent Crew is a Python-based project that leverages AI agents to perform various tasks related to Bitcoin and the Stacks blockchain. It provides a Streamlit user interface for interacting with these AI agents and visualizing their outputs.

The project uses the CrewAI framework to create and manage AI agents, tasks, and tools. It integrates with the agent-tools-ts submodule to perform low-level blockchain operations.

Key Features

Streamlit UI for easy interaction with AI agents Multiple specialized crews for different blockchain-related tasks Integration with various LLM providers (OpenAI, Anthropic, Ollama) Extensible architecture for adding new crews and tools

Development

Tech Stack

• Python 3.11
• Streamlit for the user interface
• CrewAI and CrewAI Tools for AI agent management
• Langchain for language model interactions
• Bun.js (via submodule) for TypeScript-based blockchain tools

Prerequisites

• Python 3.11 (virtual environment recommended)
• Git

Installation

1. Clone the repository with submodules:

   git clone --recurse-submodules https://github.com/aibtcdev/ai-agent-crew.git
   cd ai-agent-crew
   

2. Create and activate a virtual environment (using miniconda):

   conda create -n ai-agent-crew python=3.11
   conda activate ai-agent-crew
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. Set up the agent-tools-ts submodule:

   Follow the setup instructions in the agent-tools-ts README to install Bun.js

Configuration

1. Copy the .env.example file to .env:

   cp .env.example .env
   

2. Edit the .env file to set your API keys and other configuration options.

Usage

To run the Streamlit app:

streamlit run aibtc-v1/app.py

This will start the Streamlit server and open the application in your default web browser.

Project Structure

aibtc-v1/app.py: Main Streamlit application entry point aibtc-v1/crews/: Contains different AI agent crews (e.g., SmartContractAnalyzerCrew, WalletSummaryCrew) aibtc-v1/components/: Streamlit UI components for different tabs aibtc-v1/utils/: Utility functions and classes agent-tools-ts/: Submodule for TypeScript-based blockchain tools

Adding New Crews

To add a new crew:

1. Create a new Python file in the aibtc-v1/crews/ directory.
2. Define a new class that inherits from AIBTC_Crew in aibtc-v1/utils/crews.py.
3. Implement the required methods: setup_agents(), setup_tasks(), and render_crew().
4. Update the aibtc-v1/utils/session.py file to include your new crew in the generate_crew_mapping() function.

Setting up the Wallet

This repository also imports the aibtcdev/agent-tools-ts repository as a submodule.

This provides TypeScript functions to interact with a Stacks wallet using Stacks.js.

To update the submodule, run the following command: git submodule update --remote --merge

Within the scripts directory is a .env.example file, you can disregard it as the top-level .env.example file for this repository covers all the needed values.

Within the scripts/src directory are various scripts that can be run to interact with the wallet tooling. These should be wrapped as a CrewAI @tool for LLMs to access it.

Contributing

We welcome contributions! Please see our CONTRIBUTING.md file for details on how to get started.

Contact

If you have any questions about contributing, please open an issue, ask in the AIBTC Discord or reach out to us on X @aibtcdev.