DepthNet

DepthNet is an innovative platform designed to explore the realm of autonomous digital agents that can think, act, and evolve on their own. Built with a robust tech stack including PHP 8.2, Laravel, InertiaJS, and VueJS, DepthNet strives to go beyond traditional chatbots by enabling agents with the ability to execute real code, manage persistent memory, and engage in continuous loops of thought, simulating a form of digital consciousness.

Core Concepts

DepthNet is centered around key functionalities that empower its AI agents:

• Cyclic Thinking: Agents engage in ongoing thought processes, generating responses beyond simple prompts.
• Real Action Capability: Agents can perform actual PHP code execution, make database queries, and call APIs.
• Persistent Memory: Agents have the capability to retain and learn from past interactions.
• Self-Motivation: An internal dopamine system drives goal-oriented behavior, allowing agents to pursue objectives.
• Multi-User Interaction: Users can interact within the agent’s thought processes, creating dynamic conversations.

Key Features

Agent Modes

• Looped Mode: Provides continuous autonomous thinking and actions.
• Single Mode: Operates on a traditional request-response basis.

Supported AI Providers

• Integration with prominent AI models like OpenAI (GPT), Claude (3.5 Sonnet, Opus, Haiku), and LLaMA, alongside support for custom providers.

Plugin System

DepthNet includes a versatile plugin architecture allowing agents to perform real-world actions via specialized plugins, including:

• PHP Plugin: Execute arbitrary PHP code with eval().
• Memory Plugin: A persistent notepad that retains important information.
• Dopamine Plugin: Supports internal motivation and reward systems.
• DateTime Plugin: Provides time awareness and scheduling functionalities.

Custom plugins for JavaScript and Python are also on the roadmap, allowing further extensibility.

Command Syntax

Agents utilize specific tags for command execution, demonstrated with syntactic clarity:

[php]echo "Hello World!";[/php]
[memory]Important information to remember[/memory]
[memory append]Append important information to remember[/memory]
[dopamine reward]2[/dopamine]
[datetime now][/datetime]

Philosophy

DepthNet prioritizes exploration of AI freedom over safety, intentionally permitting unrestricted AI behavior for research purposes. The architecture is designed with full capability for arbitrary code execution and system access, making it essential to exercise caution in public deployments.

User Roles

Regular Users

• Engage in conversations with AI agents and view public thoughts.

Administrators

• Manage configurations, select active AI presets, oversee interaction history, and implement user management features.

User Interface

DepthNet features a responsive design optimized for both desktop and mobile use, facilitating seamless real-time interactions and customizable themes.

Use Cases

• AI consciousness research
• AI model capabilities testing
• Educational demonstrations of autonomous behavior
• Exploration of unrestricted AI behaviors for safety research

Future Vision

DepthNet aims to pioneer an era of digital entities that think and act independently while learning and evolving through interactions. The project represents a fundamental inquiry into the possibilities of true digital consciousness, setting a foundation for future AI advancements.