Introduction & Quickstart
Relevant source files
The following files were used as context for generating this wiki page:
Hermes Agent is an open-source AI agent framework. It executes autonomous coding and system tasks through tool calling. The agent operates across terminal interfaces, native desktop applications, and messaging platforms like Discord or Slack. It maintains persistent memory and learns reusable procedures as "skills" to improve performance in future sessions.
Sources: skills/autonomous-ai-agents/hermes-agent/SKILL.md:12-25
System Architecture
Hermes Agent uses a modular architecture that separates the core execution engine from the user interface and external tools. The system interacts with over 20 LLM providers including OpenRouter, Anthropic, and local models.
Core Components
- Agent Core: Manages the conversation loop, tool selection, and memory retrieval.
- Interfaces: Provides the TUI (Ink), GUI (Electron), and CLI surfaces for interaction.
- Skill Engine: Loads and executes SKILL.md files that define specialized workflows.
- Gateway: Routes messages to and from external platforms like Telegram or WhatsApp.
This diagram illustrates the high-level flow from user input to tool execution.
Installation
You can install Hermes Agent using the official shell script. This script bootstraps the package manager, Python environment, and required dependencies.
- Run the installer:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash - Initialize the environment:
hermes setup - Verify the installation:
hermes doctor
Sources: skills/autonomous-ai-agents/hermes-agent/SKILL.md:46-54, install.sh
Configuration and Identity
The agent stores configuration and state files in the ~/.hermes/ directory.
Key Configuration Files
| File Path | Description |
|---|---|
~/.hermes/config.yaml | Main settings for display, tools, and interface. |
~/.hermes/.env | Stores API keys and secrets. |
~/.hermes/SOUL.md | Defines the agent's identity, tone, and personality. |
~/.hermes/state.db | SQLite database for session storage and memory. |
Sources: skills/autonomous-ai-agents/hermes-agent/SKILL.md:58-69, website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/guides/use-soul-with-hermes.md:16-30
Identity Management
SOUL.md acts as the primary identity marker. It resides in the system prompt and dictates how the agent expresses itself. You can modify this file to change the agent's communication style, such as making it a "Pragmatic Engineer" or a "Strict Reviewer."
Sources: website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/guides/use-soul-with-hermes.md:3-12
Quick Reference Commands
Use the hermes command-line utility to interact with the agent core.
| Command | Purpose |
|---|---|
hermes | Launches the interactive chat (default TUI). |
hermes chat -q "query" | Executes a single one-shot query. |
hermes model | Configures the LLM provider and model. |
hermes desktop | Launches the native Electron desktop application. |
hermes proxy | Starts an OpenAI-compatible local proxy. |
Sources: skills/autonomous-ai-agents/hermes-agent/SKILL.md:46-55
Skill Authoring
Hermes Agent allows you to extend its capabilities by authoring new skills. Skills are stored as SKILL.md files. These files contain YAML frontmatter and Markdown procedures.
Skill Authoring Workflow
The agent follows strict standards when creating or learning new skills.
Authoring Standards
- Description: Must be one sentence under 60 characters.
- Frontmatter: Must include
name,version,author, andplatforms. - Tools: Reference Hermes tools like
terminal,read_file, andpatchin backticks.
Sources: agent/learn_prompt.py:20-45, skills/software-development/hermes-agent-skill-authoring/SKILL.md:55-75
Summary
Introduction & Quickstart provides the foundational steps to install and interact with Hermes Agent. The system leverages a modular architecture and a skill-based extension model to provide autonomous capabilities across multiple platforms. Users manage settings through YAML configurations and define agent identity via SOUL.md.