# Hermes Agent

Model-agnostic agent system with custom bots, Kanban, skills, hooks, and memory. Available through desktop, CLI, and messaging gateways.

Canonical profile: https://agenten-kompass.mmeierhoff.chatgpt.site/en/agents/hermes/
Editor: [Michael Meierhoff](https://agenten-kompass.mmeierhoff.chatgpt.site/en/about/)
Sources reviewed: 2026-09-30
Basis: provider documentation. Examples are editorial suggestions, not hands-on product tests.

## Overview

Hermes is a model-agnostic agent system. Its desktop app, CLI, and messaging gateway provide access. In Bot Mode, you configure multiple specialized agent profiles.

- Provider: Nous Research
- Product family: Hermes
- Profile type: Self-hosted agent
- Environment: Your own computer / server; desktop app, CLI, and chat channels
- Connections: Web, browser, and messaging channels
- Setup: Installation and model access
- Pricing model: Hosting + model usage

## Key features

### Set up your own bots

Bots receive their own roles, models, memories, and approved tools. Existing profiles can be copied as a starting point.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/bot-mode

### Bot groups and mentions

Multiple bots can collaborate in group chats and communicate directly with one another. Address specialists by name.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/bot-mode

### Kanban for multiple agents

A persistent task board manages responsibilities, dependencies, handoffs, and reviews. Tasks remain available beyond individual agent runs.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban

### Create and reuse skills

Recurring procedures are saved as Skills. Hermes can create and improve Skills while working; the Skills Hub offers additional templates.

Source: https://hermes-agent.nousresearch.com/docs/

### Hooks and plugins

Events such as tool calls or session changes can trigger configured hooks. Depending on the hook, logging, checks, or adjustments are possible.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/hooks

### Memory across sessions

Saved memories and searches of earlier sessions help reuse prior knowledge. Profiles retain their own context.

Source: https://hermes-agent.nousresearch.com/docs/

### Schedules and routines

Recurring tasks can run on a schedule. Results can be delivered through configured communication channels.

Source: https://hermes-agent.nousresearch.com/docs/

### Messaging, browser, and MCP

Hermes provides messaging gateways, web tools, and MCP connections. Configuration and access credentials determine which services are available.

Source: https://hermes-agent.nousresearch.com/docs/

### Choose models and runtime environments

You select the model provider and execution environment. Running it yourself also includes maintaining access, runtime, and tool permissions.

Source: https://hermes-agent.nousresearch.com/docs/

## Requirements and limits

### Kanban requires a running service

Requirement: The gateway service distributes open tasks to agents. They remain pending unless the gateway is running.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban

### Task board and tool permissions are separate

Requirement: The desktop board displays tasks. Its Kanban tool must also be enabled for a chat agent to manage them.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban

### Bot messages do not interrupt ongoing work

Limitation: A message to a busy bot is queued and processed later. An acknowledgement does not mean the result is complete.

Source: https://hermes-agent.nousresearch.com/docs/user-guide/bot-mode

## Example workflows

### Research with a bot team

- Your input: Topic, source scope, and separate roles for research and review.
- Possible result: A source report with Kanban tasks, handoffs, and open questions.
- What to check: Open evidence; define responsibilities and approvals before execution.

### Prepare a weekly work report

- Your input: Approved project files, a reporting skill, and a schedule.
- Possible result: A recurring draft with changes and open tasks.
- What to check: Review recency, recipients, and triggered actions.

## Context

You are responsible for installation, model access, operation, and permissions. Start with a limited test environment.

The features are documented. Bot Mode, Kanban, and hooks require a suitable installation and custom configuration.

## Sources

- https://hermes-agent.nousresearch.com/

- https://hermes-agent.nousresearch.com/docs/

- https://hermes-agent.nousresearch.com/docs/user-guide/bot-mode

- https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban

- https://hermes-agent.nousresearch.com/docs/user-guide/features/hooks

[Sources and method](https://agenten-kompass.mmeierhoff.chatgpt.site/en/sources/)