Skip to content
Hermes Agent
On Trove · from Marudi

Hermes Agent

The Complete Developer Guide

A practical, code-first developer guide to Hermes Agent, an open-source autonomous AI agent framework built by Nous Research.

You can read the contents and the first 2,000 characters.
Free accounts get the whole book.

  • By Marudi
  • 37 chapters
  • Updated Nov 2, 2026
  • 34 readers

What’s inside

  • What Is Hermes Agent?
  • The Hermes Model Family
  • Architecture Overview
  • Installation & Setup
  • Provider Configuration
  • The CLI & TUI

What is Hermes Agent?

A practical, code-first developer guide to Hermes Agent, an open-source autonomous AI agent framework built by Nous Research. Covers installation, tool usage, memory systems, skills creation, structured output, delegation patterns, multi-platform deployment, and production best practices.

Read more

Marudi · markdown · Listed Jun 8, 2026

From the first page

Author: Marudi Version: 1.0 — June 2026 Last updated: 2026-06-14 — current to Hermes Agent v2026.6.5 and model Hermes-4.3-36B.

Audience: Public marketplace reference. Maintainer / custodian: Marudi.

Scope of validity: Hermes Agent release scheme is date-based; this edition is current as of Hermes Agent v2026.6.5 (June 6, 2026) and Hermes 4.3-36B (the latest model, December 2025). Refresh cadence: review every 4–6 weeks — the framework ships multiple releases per month. (Note: releases carry both a date-based version (primary, e.g. v2026.6.5) and a semver alias (e.g. v0.16.0); secondary sources sometimes cite only the semver. Both refer to the same artifact. Source: https://github.com/NousResearch/hermes-agent/releases/tag/v2026.6.5)

Most AI agents are sophisticated command-line tools. You give them a task, they complete it, and you start fresh next session. Hermes is different.

Preview · 2,000 of 145,990 characters

Contents

  1. 01Hermes Agent: The Complete Developer Guidep.1
    1. Prefacep.1
  2. 02What Is Hermes Agent?p.2
    1. The Agent That Grows With Youp.2
    2. Key Differentiatorsp.2
    3. What Hermes Is NOTp.2
    4. The Ecosystemp.2
  3. 03Decision Matrixp.3
  4. 04The Hermes Model Familyp.4
    1. Hermes-2p.4
    2. Hermes-2 Prop.4
    3. Hermes-3 (August 2024)p.4
    4. Hermes-4 (August 2025)p.4
    5. Hermes-4.3 (December 2025) — current flagshipp.4
    6. Model Selection Guidep.4
  5. 05Architecture Overviewp.5
    1. The Three-Tier Designp.5
    2. The Agent Loopp.5
    3. NL → Code Entity Bridgep.5
    4. Context Compressionp.5
  6. 06Installation & Setupp.6
    1. Quick Installp.6
    2. Directory Structurep.6
    3. Initial Configurationp.6
    4. Verify the Installationp.6
    5. Agent rulesp.6
  7. 07Provider Configurationp.7
    1. All Supported Providersp.7
    2. Basic Provider Config in config.yamlp.7
    3. Switching Models at Runtimep.7
    4. Ollama Setup (Local Inference)p.7
    5. Custom / Self-Hosted Endpointsp.7
    6. Fallback Provider Chainp.7
    7. OpenRouter Advanced Routingp.7
    8. AWS Bedrockp.7
    9. +1 more section
  8. 08The CLI & TUIp.8
    1. Launching Hermesp.8
    2. Essential Slash Commandsp.8
    3. TUI Keyboard Shortcutsp.8
    4. Session Managementp.8
    5. Hermes Desktop (native app)p.8
See all 37 parts
  1. 09Your First Tool Usep.9
    1. Verifying Tools Are Enabledp.9
    2. First Terminal Tool Callp.9
    3. First Web Searchp.9
    4. Watching Tool Progressp.9
    5. Your First Multi-Tool Chainp.9
  2. 10Memory Systemp.10
    1. System Prompt Construction Orderp.10
    2. SOUL.md — Persona & Identityp.10
    3. MEMORY.md — Environmental Factsp.10
    4. USER.md — User Profilep.10
    5. Memory Management Commandsp.10
    6. SQLite FTS5 Session Searchp.10
    7. Memory vs. Skills: The Distinctionp.10
    8. The Reflection / Self-Evolution Loopp.10
    9. +1 more section
  3. 11Skills Systemp.11
    1. What a Skill Doesp.11
    2. SKILL.md Formatp.11
    3. Directory Structurep.11
    4. Progressive Disclosure (Token Efficiency)p.11
    5. Skill Bundlesp.11
    6. Autonomous Skill Creationp.11
    7. Installing Skills from External Sourcesp.11
  4. 12The Tool Ecosystemp.12
    1. Built-in Tools (47 total)p.12
    2. Enabling and Disabling Toolsp.12
    3. The execute_code Power Toolp.12
    4. Terminal Backendsp.12
  5. 13The Nous Tool Gatewayp.13
    1. What the Gateway Coversp.13
    2. Setupp.13
    3. Image Generation Modelsp.13
    4. Using Tools via the Gatewayp.13
  6. 14ChatML & the Hermes Prompt Formatp.14
    1. ChatML Formatp.14
    2. XML Tag Referencep.14
    3. The Canonical Function-Calling System Promptp.14
    4. Hermes-3 Scratch Pad (GOAP Reasoning)p.14
    5. Hermes-4 Hybrid Reasoningp.14
  7. 15Tool Definitions with JSON Schemap.15
    1. The Tool Definition Formatp.15
    2. Pydantic Model Classesp.15
    3. Complete Tool List Examplep.15
    4. LangChain Tool Integrationp.15
    5. Writing Effective Tool Descriptionsp.15
    6. Agent rulesp.15
  8. 16The Tool Call / Response Cyclep.16
    1. The Full Lifecyclep.16
    2. Parsing Tool Calls in Pythonp.16
    3. Full Agent Loop with Ollamap.16
    4. Recursive Function Call Loop with Depth Limitp.16
    5. Multi-Tool Calls in One Turnp.16
  9. 17Structured Output and JSON Modep.17
    1. System Prompt for JSON Modep.17
    2. Pydantic-Based Structured Outputp.17
    3. Complete JSON Mode Examplep.17
    4. Complex Nested Schema Examplep.17
    5. Structured Extraction from Documentsp.17
  10. 18Multi-Step Tool Chainsp.18
    1. When to Chain vs. When to Nestp.18
    2. Practical Chain: Research + Writep.18
    3. Using execute_code for Long Chainsp.18
    4. Handling Tool Errors Gracefullyp.18
  11. 19Delegation & Subagentsp.19
    1. delegate_task — The Built-in Patternp.19
    2. delegate_task Characteristicsp.19
    3. Spawning Independent Hermes Processesp.19
    4. Choosing the Right Patternp.19
    5. Five Proven Delegation Patternsp.19
    6. Kanban Multi-Agent Coordination (v0.13+)p.19
    7. Agent rulesp.19
  12. 20Scheduling & Cron Automationp.20
    1. Schedule Format Supportp.20
    2. Creating Jobsp.20
    3. Job Lifecyclep.20
    4. Delivery Routingp.20
    5. Cron Execution Detailsp.20
    6. Advanced Cron Optionsp.20
    7. Real-World Cron Recipesp.20
    8. Recipe: A Deterministic Scraping Watchdog (`--script` + `--no-agent` + `--deliver`)p.20
    9. +1 more section
  13. 21Custom Plugins & Toolsp.21
    1. Plugin Directory Structurep.21
    2. plugin.yamlp.21
    3. __init__.py — Full Registrationp.21
    4. tools.py — Tool Handlerp.21
    5. Available Hook Eventsp.21
    6. Plugin Discovery Pathsp.21
    7. Enabling a Pluginp.21
    8. Complete Practical Example: Slack Notifier Pluginp.21
  14. 22MCP Integrationp.22
    1. What MCP Doesp.22
    2. Basic MCP Configurationp.22
    3. Tool Filteringp.22
    4. mTLS for Secure Remote Serversp.22
    5. MCP Sampling (Server Requests LLM Inference)p.22
    6. MCP CLI Commandsp.22
    7. Running Hermes AS an MCP Serverp.22
    8. Popular MCP Servers to Installp.22
  15. 23Terminal Backendsp.23
    1. Backend Comparisonp.23
    2. Setting the Backendp.23
    3. Docker Backend Detailsp.23
    4. SSH Backendp.23
  16. 24The Messaging Gatewayp.24
    1. Supported Platforms (24+)p.24
    2. Gateway Setupp.24
    3. Configuration Filesp.24
    4. Per-Platform Environment Variablesp.24
    5. Access Tiersp.24
    6. Session Reset Policiesp.24
    7. In-Chat Slash Commandsp.24
    8. Finding a Delivery Target (JIDs & WhatsApp Groups)p.24
    9. +2 more sections
  17. 25Docker & VPS Deploymentp.25
    1. Basic Docker Gatewayp.25
    2. Full Docker Composep.25
    3. Resource Requirementsp.25
    4. VPS Warningp.25
    5. S6 Supervisionp.25
    6. Multi-Tenant Isolationp.25
  18. 26Serverless Deploymentp.26
    1. Modalp.26
    2. Daytonap.26
    3. Cost Patternsp.26
  19. 27Security & Command Approvalp.27
    1. Approval Modesp.27
    2. The Approval Promptp.27
    3. Hardline Blocklist (Never Overridable)p.27
    4. Dangerous Pattern Categoriesp.27
    5. YOLO Modep.27
    6. Seven-Layer Defense Modelp.27
    7. Production Security Checklistp.27
    8. Bitwarden Secrets Manager Integrationp.27
    9. +2 more sections
  20. 28Prompting Hermes Effectivelyp.28
    1. Give Rich Upfront Contextp.28
    2. Use AGENTS.md for Project Contextp.28
    3. Let the Agent Explorep.28
    4. Use Skills for Recurring Workflowsp.28
    5. Prompt Cachingp.28
    6. Memory Is a Frozen Snapshotp.28
    7. The most Powerful Turnp.28
    8. Agent rulesp.28
  21. 29Production Patternsp.29
    1. The Agent-Loop Internals (Developer View)p.29
    2. Model Flexibilityp.29
    3. Multi-Profile Setupp.29
    4. Error Handling in Productionp.29
    5. Structured Loggingp.29
  22. 30Local Models with Ollama & LM Studiop.30
    1. Choosing Local Models for Agentsp.30
    2. Ollama Setup (Complete)p.30
    3. LM Studio Setupp.30
    4. Why Context Length Mattersp.30
    5. Agent rulesp.30
  23. 31Complete Configuration Referencep.31
    1. Full config.yamlp.31
    2. Configuration Management Commandsp.31
  24. 32Slash Command Referencep.32
    1. Session Commandsp.32
    2. Model & Configurationp.32
    3. Background & Asyncp.32
    4. Skillsp.32
    5. Schedulingp.32
    6. Platformp.32
    7. Voicep.32
    8. Status & Diagnosticsp.32
    9. +1 more section
  25. 33Provider & Model Selection Guidep.33
    1. Decision Frameworkp.33
    2. For Hermes Agent (Framework)p.33
    3. For Direct Hermes Model Use (API / Ollama)p.33
    4. Context Length Requirementsp.33
  26. 34Migration from OpenClawp.34
    1. Migration Commandp.34
    2. What Migratesp.34
    3. Known Issuep.34
  27. 35Appendix A: Quick Start Cheat Sheetp.35
  28. 36Appendix B: Tool Call Templatep.36
  29. 37Appendix C: Environment Variables Referencep.37
    1. About This Guidep.37
    2. ACP Server Integration — Protocol Roles Clarificationp.37
    3. WhatsApp Group Messaging — Discovery and Pairing Correctionsp.37
    4. Gateway & Dashboard Bind Address Configurationp.37
    5. The `pre_gateway_dispatch` Hook: Payload and Contractp.37