Set up and manage AI agents, chat, and tools from a desktop interface
A window opens where you can configure, chat with, and manage the memory and tools of an AI assistant without typing complex commands in a terminal. This app installs Hermes Agent, a self-improving AI assistant, on your desktop and provides a graphical interface to connect various AI models and messengers. It hides complex backend settings and guides you to select only the features you need.
Automatic installation and setup on first run
When you first open the app, it checks if Hermes Agent is already installed. If not, it runs the official installation script to automatically resolve necessary dependencies. You choose whether to run it locally or connect to a remote server, and specify the AI provider or local model endpoint to use. Once this process is complete, the main workspace opens, handling chat, sessions, profiles, memory, tools, scheduling, and messaging gateways. All settings are stored and managed in the ~/.hermes folder.
Model providers and chat features
It supports various AI providers such as OpenRouter, Anthropic, OpenAI, Google Gemini, xAI Grok, Nous Portal, Qwen, MiniMax, Hugging Face, and Groq. Local model servers like LM Studio, Ollama, vLLM, and llama.cpp can also be connected as OpenAI-compatible endpoints. In the chat screen, you can view token usage and costs in real time via SSE streaming, and see responses with markdown rendering and syntax highlighting. You can quickly perform tasks like creating new sessions, web search, image generation, code execution, and memory management using 22 slash commands.
Messaging gateway and automation
You can configure 16 messaging platforms as gateways, including Telegram, Discord, Slack, WhatsApp, Signal, Matrix, Mattermost, email, SMS, iMessage, DingTalk, Feishu, WeCom, WeChat, Webhooks, and Home Assistant. Using the cron job builder, you can create automation tasks with various cycles from minute-level to custom cron, and send results to 15 delivery targets. It also includes a feature to manage development servers and adapters through a 3D visual interface called Hermes Office (Claw3d).
API key and secret management
By default, API keys are stored in the ~/.hermes/.env file. If you do not want to store keys in a plaintext file, you can integrate with an external secret manager through a command provider. It supports various tools such as KeePassXC, GnuPG, pass, secret-tool, Bitwarden, and 1Password, reading values from the standard output of helper commands. This method does not require hardware security modules like TPM or FIDO2, and applies a 3-second timeout and 1MiB output limit to helper execution to prevent the app from hanging. Resolved values are not logged or written to disk.
Platform-specific installation notes
This project is community-maintained and in an active development stage, so features may change or some may break. Windows users will see a SmartScreen warning because the installer lacks code signing, requiring them to click 'More info' to run it. Fedora users need the --nogpgcheck option because the .rpm file lacks a GPG signature, and automatic updates are not supported, requiring manual reinstallation for new versions. WSL users may encounter a request for a sudo password during Playwright installation, necessitating a temporary grant of passwordless sudo privileges followed by removal. This app depends on the Hermes Agent project, and if using a local model provider, that server must already be running.