Run a self-hosted AI assistant that remembers your project context across sessions so you never have to re-explain your setup.
Schedule recurring tasks for your AI agent to run automatically even when you are not at your computer.
Chat with your Hermes agent from Telegram, Discord, or Slack on your phone without opening a browser.
Hand off complex coding tasks to Claude Code from inside Hermes and have results saved back to the agent's memory automatically.
Requires an AI provider API key (OpenAI, Anthropic, Google, DeepSeek, or others), the setup script creates the Python environment and opens the browser automatically.
Hermes is an AI agent that runs on your own computer or server and remembers what it learns across sessions. Most AI chat tools start fresh every time you open them, forgetting your preferences, your projects, and how you like to work. Hermes keeps all of that in memory, grows more capable the longer you use it, and can run scheduled tasks for you even when you are not at your computer. Hermes WebUI is a browser-based interface for that agent. Instead of typing commands in a terminal, you get a three-panel layout: a sidebar on the left for switching between chat sessions, a center panel for the conversation itself, and a right panel for browsing files in your workspace. A small ring icon shows how much of the model's context window you are using. Controls for picking a model, switching profiles, and adjusting settings live at the bottom of the chat input area. The interface is built with plain Python on the server side and ordinary JavaScript in the browser, with no build tools, no bundling, and no framework to install. You start it with a single command, and the setup script handles detecting whether the Hermes agent is already installed, creating a Python environment, and opening the browser for you. For people running it on a home server or cloud VM, there is also a control script that manages it as a background process with start, stop, restart, and log commands. A Docker image is available for those who prefer containers. Beyond the web interface, Hermes itself supports connecting from messaging apps such as Telegram, Discord, Slack, and Signal, so you can interact with your agent from a phone without opening a browser. It can write and save its own reusable procedures based on what it learns, works with models from OpenAI, Anthropic, Google, DeepSeek, and others, and can hand off heavy coding tasks to tools like Claude Code and bring the results back into its own memory. It is open source under the MIT license and entirely self-hosted, meaning your conversations and memory stay on your own hardware. The full README is longer than what was shown.
← nesquena on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.