Vyrtuous

Vyrtuous is a Java-based application designed to integrate local and remote AI tools.

Project Flow

Project Description

This project aims to provide a unified interface for interacting with various AI models and services, including local LLMs and cloud-based APIs like OpenAI, OpenRouter, and LM Studio. It supports features like context management, approval workflows, and a REPL interface for experimentation.

Key Features

• Integration with Multiple AI Tools: Vyrtuous integrates with Discord and provides a framework for connecting to various AI models and services. Currently, the codebase demonstrates Discord integration.
• Context Management: Maintains and manages conversation context for more coherent interactions.
• Approval Workflows: Implements approval mechanisms to control AI responses.
• REPL Interface: Provides a Read-Eval-Print Loop (REPL) for interactive experimentation with AI tools.
• Modular Design: Built with modularity in mind, allowing for easy extension and customization.

Technologies Used

• Asynchronous Programming: CompletableFuture for non-blocking operations.
• Discord API (JDA): For bot integration.
• Java: The primary programming language.
• Logging: Utilizes Java's logging framework.
• Maven: The primary package manager (Gradle support upcoming).
• Model Context Protocol Server: A backend component to reveal context to the LLM's.
• Spring-Boot-AI: Autowires beans, component, services and the main SpringBootApplication.

Setting Up the Project

This project is built using Maven.

Prerequisites:

• Java 17 or higher installed and configured.
• Maven installed and configured.

Installation (Maven):

1. Clone the repository.
2. Run mvn clean install to build the project and download dependencies.
3. Copy Vyrtuous-0.1.jar to the root directory of the project as Vyrtuous.jar.
4. Configure the application with appropriate settings in .env and save it to ~/.env or add the fields to your .env file.
5. Linux: docker build --platform=linux/arm64/v8 -t vyrtuous .
6. Mac: docker buildx build --platform=linux/arm64/v8 -t vyrtuous --load .
7. docker run -it --env-file ~/.env --rm vyrtuous

Configuration:

1. Configure API keys, model paths, and other settings in the appropriate configuration files (e.g., .bashrc). Refer to the project's documentation for specific details.

Description

Dependencies:

This project relies on the following key dependencies (listed in pom.xml):

• Apache Http Client
• Jackson Databind
• JDA
• Metadata
• Spring-Boot

** Package Structure:**

• package com.brandongcobb.vyrtuous.service;: This package is for services, indicating their role in servicing requests from other parts of the code.
• package com.brandongcobb.vyrtuous.objects: This package is for objects, indicating their role in serializing data from AIService.
• package com.brandongcobb.vytuous.tools: This package is for tools, self explanatory.
• package com.brandongcobb.vyrtuous.utils.handlers: This package is for utilities accessing the serialized data in the program's objects.
• package com.brandongcobb.vyrtuous.utils.inc: This package is for information which doesn't have a better place.
• package com.brandongcobb.vyrtuous.record: This package is for recording model and tool statistics.
• package com.brandongcobb.vyrtuous.cogs: This package has two functions and are registered under DiscordBot; one exposes commands to Discord and the other exposes a message listener.
• package com.brandongcobb.vyrtuous.enums: This package is critical to the function of the program and also pose a vulnerability returning unserializable data from the models.
• package com.brandongcobb.vyrtuous.domain: This package formalizes the connection between JSON tool schemas and their programatic fields.
• package com.brandongcobb.vyrtuous.component: This package contains the Discord bot and the Model Context Protocol Server.

Model Context Protocol Server

The server is a core component of Vyrtuous, responsible for managing model contexts and providing a central interface for interacting with various AI tools. It acts as a server that handles requests and orchestrates the execution of different tools.

• count_file_lines: Counts the number of lines in a file.
• create_file: Creates a new file with specified content.
• find_in_file: Searches for specified terms within a file, providing context lines around matches.
• patch: Applies patches (replacements, insertions, deletions) to files.
• read_file: Reads the contents of a file.
• search_files: Searches for files matching specified criteria.
• search_web: Searches the web for information using the Google Programmable Search API.

Advanced Features

Vyrtuous now includes a Model Context Protocol Server and cloud access to OpenAI:

• OpenAI: The leader in global AI can be accessed with a single API key and personal funding.
• Model Context Protocol Server: The opportunities are vast and the tools only reflect a small portion of useful actions for agentic capabilities.
• Discord: Local LLM's with streaming is highly recommended.

Future Enhancements

• Further implementation of Spring-Boot AI
• Native support for tool calling in local models.
• Refactor codebase for improved scalability and maintainability.

License

This project is licensed under the GNU General Public License v3.