Self-hosted AI chat
for your Hermes agent
Hermium gives you a clean, fast chat interface that connects to your Hermes Agent. One command to install. Zero config to run.
Free & open source.
Everything you need to talk to your agent
Hermium is a clean, fast chat interface built specifically for the Hermes agent. No fluff, no lock-in.
Self-hosted. Your data never leaves.
Run everything locally. No cloud accounts, no telemetry, no external APIs phoning home. Just you, your machine, and your agent.
const [data, setData] = useState();
};
Chat with markdown, code, and streaming
Full markdown rendering with syntax-highlighted code blocks, streaming responses in real-time, and keyboard shortcuts for power users.
Drop files into the chat
Attach any file — code, documents, images — and the agent reads them inline. It can edit, reference, or execute code from what you upload.
Switch models mid-conversation
Change the active model without starting over. Toggle reasoning depth from off to xhigh when you need deep analysis or just a quick answer.
Persistent sessions that remember
Every conversation is saved automatically. Rename sessions, pin the important ones, and pick up anywhere. Agent memory survives across chats.
$ hermium start
→ Server running at localhost:42424
One command to install
npm install -g hermium && hermium start. That's it. The API and web UI start together. Zero configuration, zero dependencies beyond Bun.
Install in seconds
One command. Zero configuration.
npm install -g hermium && hermium startInstall
Run the command above with Bun or npm.
Start
hermium start launches both API and web UI.
Chat
Open http://localhost:42424 and start talking.
Requires a running Hermes Agent.