How to Install and Set Up Hermes Agent (NousResearch) from GitHub — Step-by-Step Guide (2026)

In the rapidly evolving landscape of autonomous AI, Nous Research’s Hermes Agent has emerged as a powerhouse for users seeking a self-improving, open-source assistant. Released in February 2026, Hermes Agent isn’t just another chatbot; it is a persistent entity that learns from every interaction, creates its own skills, and bridges the gap between local terminal power and multi-platform accessibility. As of April 2026, with the release of version 0.10.0, Hermes has matured into a production-ready framework that supports over 200 LLM models via OpenRouter and Nous Portal, features a sophisticated SQLite-backed memory system, and integrates with more than 15 messaging platforms.

Setting up a state-of-the-art AI agent used to require complex environment configurations and hours of debugging. Hermes Agent changes this narrative with a streamlined, “one-click” philosophy. Whether you are a developer looking to automate your GitHub workflows or an SMB owner seeking to integrate AI into your Slack or Telegram channels without a massive engineering overhead, Hermes offers a robust foundation. This guide provides a comprehensive walkthrough for installing and configuring Hermes Agent, ensuring you can go from a fresh terminal to your first agentic conversation in minutes.

Understanding the Hermes Agent architecture

Before diving into the installation, it is essential to understand what makes Hermes Agent unique. Unlike traditional LLM wrappers, Hermes operates on a “closed learning loop.” It maintains persistent memory through a multi-layer system: a short-term MEMORY.md for immediate context, a USER.md profile for long-term preferences, and a deep SQLite database with FTS5 full-text search for cross-session recall. This allows the agent to remember your specific coding style, previous project details, and even the “pitfalls” it encountered during past tasks.

The agent is built primarily in Python (87.7%) and leverages the uv package manager for lightning-fast dependency resolution. It supports six distinct terminal backends, including local execution, Docker, SSH, and serverless options like Modal. This flexibility means you can run your agent on a $5 VPS or a massive GPU cluster, with the same consistent interface across CLI, Telegram, Discord, Slack, and WhatsApp.

Prerequisites and system requirements

As of April 2026, Hermes Agent is optimized for Linux, macOS, and WSL2. While native Windows is not supported, WSL2 provides a seamless experience. The beauty of the Hermes installer is its “zero-manual-dependency” approach; the only tool you strictly need to have installed is Git.

For mobile enthusiasts, Hermes also supports Android via Termux. The installer automatically detects the environment and applies a specific .[termux] extra that skips heavy voice dependencies while maintaining full agentic functionality. This makes Hermes one of the most portable open-source agents available in 2026.

Step 1: The one-command installation

The most efficient way to install Hermes Agent is via the official curl script. This script automates the cloning of the GitHub repository, sets up the virtual environment using uv, and configures the hermes CLI command globally.

1. Open your terminal (or WSL2).
2. Run the following command:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Once the script completes, you must reload your shell to activate the hermes command. You can do this by running source ~/.bashrc (or source ~/.zshrc for macOS users). You can verify the installation by typing hermes version, which should display the latest version—v0.10.0 (released April 16, 2026).

Step 2: Configuring models and tools

With the binary installed, the next step is to run the interactive setup wizard. While you can set individual keys using hermes config set, the wizard is the recommended path for new users as it ensures all critical components are configured correctly.

hermes setup

The wizard will prompt you for three main things:

• LLM Provider: Choose your primary provider. OpenRouter is highly recommended for access to 200+ models, but you can also use the native Nous Portal (which currently offers a free tier for the Xiaomi MiMo v2 Pro model).
• API Keys: Paste your provider’s API key. Hermes uses secure redaction in logs to ensure your secrets never leak.
• Tool Selection: You will be presented with a list of 40+ built-in tools. These include terminal (for CLI execution), browser_use (for web navigation), and specialized skills like gitnexus-explorer.

If you are migrating from the older OpenClaw project, Hermes will detect your ~/.openclaw directory and offer to import your personas (SOUL.md), memories, and existing API keys automatically.

Step 3: Launching your first conversation

Once the setup is complete, you can start a conversation directly from your terminal. Hermes features a rich Terminal User Interface (TUI) with multiline editing and slash-command autocomplete.

hermes chat

Inside the chat, you can use several powerful slash commands to manage the agent’s behavior on the fly:

• /model [provider:model]: Switch your LLM mid-conversation (e.g., /model nous:hermes-3-pro).
• /skills: Browse and enable new procedural skills created by the community.
• /usage: Check your current token consumption and cost insights.
• /retry or /undo: Quickly correct the agent if a tool call goes off-track.

A standout feature added in v0.8.0 is notify_on_complete. You can ask Hermes to start a long-running background task—like training an AI model or running a complex test suite—and it will notify you once it finishes, allowing you to work on other tasks in the meantime without constant polling.

Step 4: Setting up the Messaging Gateway (Optional)

One of Hermes’ most compelling features is the ability to talk to your agent from anywhere. The Messaging Gateway allows you to link your agent to Telegram, Discord, Slack, and even Signal. This is why many SMBs choose to deploy Hermes on a remote VPS; it acts as a persistent office assistant that is always one message away.

1. Run hermes gateway setup and choose your preferred platform (e.g., Telegram).
2. Follow the prompts to provide a Bot Token (generated via BotFather for Telegram).
3. Start the gateway service with hermes gateway start.
4. Optionally, use hermes gateway enable-autostart to ensure the agent runs as a systemd service, surviving reboots.

In the messaging interface, you can even use Approval Buttons on Slack and Telegram. When the agent attempts a “dangerous” command (like deleting a file or sending an email), it will send you a message with “Approve” and “Deny” buttons, keeping you in control of the loop even while you’re on the go.

Conclusion

Hermes Agent represents a significant leap forward for open-source AI. By combining a one-command installation process with a sophisticated learning architecture, Nous Research has created a tool that is both accessible to beginners and powerful enough for enterprise automation. As of April 2026, its ability to persistently remember interactions and self-improve through new skills makes it a top-tier choice for developers and businesses alike. Whether you are running it locally for personal productivity or deploying it via the Messaging Gateway to empower your entire team, the steps outlined above will have your agent up and running in no time. For your next steps, explore the /skills menu to see how community-contributed workflows can further extend your agent’s capabilities.