Hermes Agent
A self-improving open-source agent that lives on your server, remembers you and writes its own skills.
Hermes Agent from Nous Research is an MIT-licensed agent that runs on your own machine, keeps persistent memory, generates skills from the work it does, and talks to you over Telegram, Discord, Slack, WhatsApp, Signal, email or the terminal. This course covers install, models, memory, skills, sandboxing and safe operation.
What you will be able to do
- Explain what Hermes Agent is, how it differs from OpenClaw and Claude Code, and when to use each
- Install Hermes on a server or workstation, choose between your own provider keys and Nous Portal, and connect a messaging platform
- Use persistent memory and auto-generated skills deliberately, and curate what the agent learns
- Choose an execution backend and use isolated subagents for parallel work
- Operate it safely: sandboxing, secrets, access control and regular review of what it has learned
Course outline
What Hermes Agent is
2 lessons · quizAn open-source, self-improving agent, and how it compares with the other tools in this track.
- An open-source, self-improving agent from Nous Research 12 min
- Hermes vs OpenClaw vs Claude Code: which for what 12 min
- Module quiz 5 questions
Install and first run
4 lessons · quizInstall Hermes, decide how it will get its models, and connect a messaging platform.
- Installing 10 min
- Models: your own keys vs Nous Portal 12 min
- Connecting a messaging platform 12 min
- Exercise: first conversation and first remembered fact exercise · 30 min
- Module quiz 5 questions
Memory, skills and subagents
5 lessons · quizHow Hermes remembers, how it writes its own skills, how it runs commands in isolation, and how it talks.
- Persistent memory and how it learns 12 min
- Auto-generated skills and curating them 14 min
- Execution backends and isolated subagents 14 min
- Voice 10 min
- Exercise: teach it one workflow and watch it reuse it exercise · 30 min
- Module quiz 5 questions
Operating it safely
3 lessons · quizSandboxing, secrets, access, and the habit of reviewing what the agent has learned.
- Sandboxing and container hardening 12 min
- Secrets, access and who can message it 12 min
- Reviewing what it learned 12 min
- Module quiz 5 questions