Hermes

Hermes is an open-source autonomous agent framework developed by Nous Research. It is not just a coding assistant inside an editor or a simple wrapper around one chat API; it is meant to act as an autonomous agent that grows with the user. It has persistent memory, so it can learn your projects and remember how problems were solved before, and it supports multiple platforms for seamless switching. It can run natural language scheduled automation for tasks like reporting, backup, and briefings, and it uses isolated sub-agents to parallelize pipelines with low context cost. It also offers sandboxed backends with container hardening and namespace isolation, plus full web and browser control. It is released under the MIT open-source license, so you can use and modify it for free.

- Automate reports, backups, and briefings on a schedule

- Run parallel tasks with isolated sub-agents

- Let it control the web and browser for you

- Keep a persistent memory of your projects and fixes

- Developers who want a real autonomous agent

- Teams automating repetitive web and file tasks

- Anyone building a personal agent under MIT terms

Because the agent can act on its own and control the browser, its actions need oversight and testing before you trust them with important work. Setup and self-hosting take real technical skill, since you connect backends and configure the sandbox yourself, and you also manage your own infrastructure and running costs. Being open-source means the code is free to change, but that also puts the burden of updates, security, and stability on you. Start with small, low-risk tasks and review what the agent does before you hand over anything sensitive.