Hermes WebUI is a lightweight, browser-based interface designed to provide full-featured access to the Hermes Agent, an advanced autonomous AI agent that operates on a user's server. The primary goal of Hermes WebUI is to deliver a seamless, user-friendly experience for interacting with Hermes Agent from any device, including desktops and mobile phones, without sacrificing any of the capabilities available in the command-line interface (CLI). The project is implemented in Python with vanilla JavaScript, deliberately avoiding heavy frameworks or build steps to ensure simplicity and ease of deployment.
The interface is organized into a three-panel layout: a left sidebar for session management and navigation, a central panel for chat interactions, and a right panel for browsing and managing workspace files. The composer footer, always visible at the bottom, provides quick access to model selection, user profiles, and workspace controls, as well as a visual indicator for token usage. All settings and session tools are consolidated in the Hermes Control Center, accessible from the sidebar, making configuration and management straightforward.
Hermes WebUI offers nearly one-to-one parity with the Hermes CLI, meaning users can perform all agent-related tasks through the web interface that they could in the terminal. This includes managing sessions, chatting with the agent, browsing and editing files in the workspace, and configuring agent settings. The WebUI is designed for secure remote access, supporting SSH tunneling for users who wish to connect from different devices or locations. The setup process is streamlined: a single command can launch the WebUI, and another can establish a secure tunnel for remote access.
A standout feature of Hermes Agent, and by extension the WebUI, is its persistent memory. Unlike many AI tools that reset context between sessions, Hermes retains user profiles, agent notes, and a library of reusable skills, allowing it to learn and adapt over time. The agent can also schedule and execute jobs while the user is offline, delivering results through various messaging platforms such as Telegram, Discord, Slack, Signal, and email. Hermes supports over ten messaging platforms, making it highly accessible from both desktop and mobile environments.
Hermes is provider-agnostic, capable of working with models from OpenAI, Anthropic, Google, DeepSeek, OpenRouter, and others. It can orchestrate other agents for specialized tasks, such as code generation, and integrate their results into its own memory. The system is entirely self-hosted, ensuring that user data and conversations remain private and under the user's control.
The repository emphasizes ease of installation and configuration. The bootstrap script automatically detects or installs Hermes Agent, sets up the necessary Python environment, and launches the web server. There is support for Linux, macOS, and WSL2, with community-maintained guides for native Windows setups. Docker deployment is also supported for both single- and multi-container environments.
Hermes WebUI is open source under the MIT license and is part of the broader Hermes ecosystem developed by Nous Research. Its focus on persistent memory, self-improving skills, broad messaging platform support, and a robust, accessible web interface distinguishes it from other agentic AI tools, making it a powerful solution for users seeking a self-hosted, always-on AI assistant.
