Hermes is an open-source agent runtime. It runs on your machine, holds memory across sessions, uses tools on your behalf, and can operate on a schedule without you watching.
This guide walks through what Hermes is, how it works, and how to design an agent team for a real workflow — from a single agent doing keyword research to a full team that researches, writes, and reviews content on a weekly cadence.
No coding experience required. Each chapter builds on the last.
Why one-off AI chats lose context, can't use tools persistently, and require you to repeat yourself every session — and what changes when your AI agent runs in a persistent environment.
A chat app responds to prompts. An agent runtime runs tools, holds memory, and operates on a schedule. The distinction matters — and a better runtime does not fix a bad workflow.
Profiles, skills, memory, toolsets, gateway, cron, subagents, and kanban — how the pieces combine into an agent that remembers, reasons, and acts across sessions.
Python 3.11, the ~/.hermes directory, model providers, tool registry, and MCP — explained in plain English for non-technical builders.
The CLI is the primary interface. Twenty-plus messaging platforms serve as input channels. A dashboard exists when you need it. Continuous vs on-demand operation modes suit different workflows.
What the install command does step by step, how to configure your first agent, and the differences between running on your laptop, a VPS, or Docker.
Three paths — local, VPS, Docker — each with trade-offs. Which one fits your situation, what can go wrong, and how to back up your work.
The agent loop: observe, think, act, repeat. How tool calls happen, when approvals kick in, and what session history does across turns.
What makes one agent better than another: narrow role design, clear instructions, skill documents, output contracts, and feedback loops.
Profiles for durable roles, subagents for one-off tasks, and kanban coordination to keep multiple agents from stepping on each other's work.
What agents store, when they retrieve it, and how memory differs from skill. Auto-learning means the agent follows your instructions to persist information — not that it independently becomes smarter.
Scheduled tasks, natural-language cron, no-agent mode, webhook triggers, and when a human still needs to approve before the next step.
The capstone: which roles become profiles, which become subagents, what goes into shared memory, and how a minimum viable setup differs from a mature one.
Secrets handling, approval workflows, provider fallback, silent cron failures, backup strategies, and safe update procedures.
Chapter 1 starts with the problem — why your current AI chat sessions lose context and can't carry work forward — and shows what changes when your agent runs in a persistent environment.
Book a Call