Hermes Agent
Hermes Agent is a tool-using AI agent that can run in terminals, messaging platforms, IDEs, and automation workflows. It is not just a chatbot: it can read and write files, execute shell commands, automate browsers, schedule jobs, search prior sessions, and delegate work to subagents.
The core vocabulary
| Term | Meaning | Examples |
|---|---|---|
| CLI command | A terminal command that controls Hermes itself | hermes doctor, hermes model, hermes skills list |
| Slash command | A command typed inside a running session | /model, /skill, /reset, /cron |
| Toolset | A group of tools that can be enabled or disabled together | terminal, file, browser, skills |
| API tool | A callable function exposed to the agent in a session | read_file, terminal, search_files, cronjob |
| Skill | A reusable procedure document for a task | debugging, GitHub workflows, Obsidian publishing |
| Profile | An isolated Hermes runtime with separate config, sessions, skills, and memory | personal, work, experiment profiles |
Why this matters
The words “command,” “tool,” and “skill” sound similar, but they operate at different layers.
- Commands control Hermes.
- Tools let Hermes act on the outside world.
- Skills teach Hermes how to perform a task reliably.
To use Hermes well, choose the model, the toolsets, and the relevant skills together.
Start here
| Goal | Starting point |
|---|---|
| Check environment health | hermes doctor, hermes status --all |
| Change models | hermes model or /model |
| Inspect tools | hermes tools list or /toolsets |
| Inspect skills | hermes skills list or /skills |
| Check the messaging gateway | hermes gateway status or /platforms |
| Check scheduled jobs | hermes cron list or /cron |