Hermes Agent: Zero to Personal AI Assistant (1 Hour Course)

Introduction to Hermes Agent

Overview of Hermes Agent

• Hermes Agent is described as a powerful AI agent that evolves with user interaction, featuring a self-improving loop for skill enhancement.

• The setup process is straightforward and does not require specialized hardware; it can be installed on a private server.

• Out of the box, Hermes comes with 684 total skills, including 91 built-in skills available immediately upon installation.

Demonstration of Capabilities

• A demonstration showcases Hermes generating voice notes and text responses, illustrating its ability to handle automation tasks (referred to as "crons").

• Examples of crons include daily news briefings and YouTube comment monitoring, highlighting its practical applications in community engagement.

Comparing AI Agents: Hermes vs. Others

Comparison with Other Tools

• The speaker plans to compare Hermes with Cloud Code and OpenClaw, discussing their respective strengths and weaknesses.

• Hermes operates on various infrastructures (e.g., Mac Mini, VPS), making it versatile compared to other agents like Claude Code which are more focused on coding tasks.

Use Cases for Different Agents

• Claude Code is used primarily for knowledge work at a desk while OpenClaw and Hermes are preferred for mobile use cases where quick task management is needed.

Key Features of Hermes Agent

Self-improvement Mechanism

• One major advantage of Hermes over competitors is its focus on self-improvement through skill creation based on user interactions.

Integration with Other Tools

• Users can integrate multiple agents (like Cloud Code and OpenClaw), allowing them to leverage different functionalities across projects seamlessly.

Understanding the Five Pillars of Hermes

Memory Management

• Memory consists of two main files: user.md (personal preferences and style) and memory.md (project context), ensuring continuity across sessions.

Skills Development

• Skills act as procedural memory or reusable playbooks that help automate tasks consistently. They are stored in skill.md files.

Personality Customization in AI Agents

Shaping the Assistant's Character

• The 'Soul' file allows users to define the personality traits of their agent, influencing how it interacts with others based on user-defined characteristics.

Automation Through Cron Jobs

Scheduled Tasks Functionality

• Cron jobs enable proactive scheduling within the agent, allowing users to set up automated tasks that run independently at specified times.

Continuous Improvement Loop

Learning from Interactions

• The self-improving loop emphasizes that continuous usage enhances the agent's understanding by saving experiences as memory or skills.

Setting Up Your Own Hermes Agent

Installation Process Overview

• The video will guide viewers through setting up the Hermes agent using a Virtual Private Server (VPS), specifically recommending Hostinger for hosting services.

How to Set Up a VPS and Manage Hermes Agents

Understanding VPS and Docker Installation

• A Virtual Private Server (VPS) provides an IP address and password for SSH access, allowing users to manage files and install applications.

• Users can install Hermes directly on the VPS at the root level or use a one-click Docker image for a containerized environment, which is simpler.

• The speaker prefers using Docker containers for managing multiple Hermes agents due to ease of installation and organization.

Managing Multiple Agents with Cloud Code

• The speaker created a cloud code project called "upit agents" to manage various VPS agents, including trading bots and personal assistants.

• This project helps keep track of passwords, environment variables, IP addresses, and configurations across different servers.

• Using cloud code simplifies management tasks as it allows users to avoid deep technical knowledge about VPS or terminal commands.

Setting Up New VPS Instances

• The speaker demonstrates setting up a new VPS instance while emphasizing that understanding complex commands isn't necessary thanks to tools like Hermes agent and Cloud Code.

• Accessing the Hostinger dashboard reveals options for managing root access, changing hostnames, viewing server analytics, and organizing multiple servers effectively.

Installing Hermes via Docker

• Two methods are available: installing at the root level through terminal commands or using the one-click install option in Docker manager.

• The speaker opts for the one-click install method for simplicity; this approach allows spinning up multiple containers within the main server.

Configuring Admin Settings

• After initiating installation via Docker manager, users must set an admin username and password for their Hermes agent during setup.

• Important credentials are saved in an organized manner within the cloud code project to maintain easy access later on.

Onboarding Process with Hermes Agent

• Once deployed, users can interact with their newly set-up Hermes agent through its command line interface (CLI).

• During onboarding, users select inference providers like OpenAI Codex GPT OAUTH instead of API keys for cost efficiency.

Connecting Telegram Bot

• Users create a Telegram bot via BotFather by providing names and usernames; this bot will be linked with their Hermes agent.

• User IDs are obtained from Telegram's user info bot to restrict access only to specific accounts interacting with the Hermes agent.

Finalizing Setup & Features Available

• After completing setup steps such as connecting user IDs and configuring channels, various tools become available including vision capabilities and task planning features.

Importance of Backup Systems

• It’s crucial to connect your Hermes agent to a GitHub repository for backup purposes; this ensures data is not lost if issues arise with the VPS.

Managing GitHub Integration Safely

• Users should generate personal access tokens carefully when integrating GitHub into their projects while ensuring sensitive information remains secure.

Creating a Classic Token and Setting Up GitHub Integration

Generating a Classic Token

• The process begins with creating a classic token instead of a fine-grained one, allowing for specific action scopes to be selected.

• Users should grant read access to public repositories and invitations while limiting write access to maintain security.

Updating the GitHub Token

• After generating the token, it is copied and set in the Hermes configuration using hermes config set GitHub token.

• This ensures that the new token is correctly integrated into the system for future operations.

Creating a Private Repository

• A private repository is created, which can be accessed through Telegram, where permissions can be adjusted (allow once, session-based, or always).

• The user opts for "always allow" to facilitate seamless daily sync operations without interruptions.

Setting Up Skills and Cron Jobs

Establishing Daily Sync Skills

• The first skill involves setting up a cron job that pushes changes to the GitHub repository every night at midnight Central Time.

• Users can also create temporary crons for short-term tasks like monitoring YouTube comments after posting videos.

Running Terminal Commands

• As skills are being established, terminal commands will run automatically to create skills and update memory as needed.

Differences Between CLI and Telegram Interfaces

Functionality Comparison

• Both interfaces utilize the same agent; however, Telegram offers less control compared to using Hermes in the terminal (CLI).

• The CLI serves as an advanced tool for deep work with better context management capabilities than Telegram's more casual interface.

Context Management Challenges

• In Telegram, users may experience ambiguity regarding session resets and context visibility due to its token-based model rather than message-based interactions.

Managing Agent Behavior and Memory

Cron Job Setup Confirmation

• Successful setup of nightly sync cron jobs is confirmed by checking file writes and memory updates within Hermes' environment.

Importance of Memory Maintenance

• Stale memory often leads to erratic agent behavior; thus regular checks on memory files are crucial for maintaining optimal performance.

Best Practices for Using Hermes Agents

Structuring Agent Accounts

• Each Hermes agent should have its own account with limited API key scopes similar to how one would manage an intern's access rights in a business setting.

Security Measures

• Implementing firewalls on VPS environments enhances security; users are encouraged to research firewall setups or consult their agents for assistance in securing their systems effectively.

Scaling Up Your Hermes Setup

Containerization Strategy

• Each agent should ideally operate within its own container on a VPS, ensuring separate keys and reducing risk from potential conflicts between agents’ functionalities.(3097)

Decision-Making Framework

• When considering whether to create new agents or expand existing ones, assess if they require distinct tools or long-term memory based on their roles.(3197)