NousResearch/hermes-agent

⭐ 141,797  ·  Python  ·  GitHub Repo

The agent that grows with you

ai ai-agent ai-agents anthropic chatgpt claude claude-code clawdbot

1-Sentence Summary

A self-improving AI agent that learns from experience, persists knowledge, and deploys anywhere without model lock-in.

🔥 Key Capabilities & USP

• Closed Learning Loop: The agent autonomously creates and refines skills from complex tasks, persists session memory with FTS5 search and LLM summarization, and builds a deepening user model via Honcho dialectic. This solves the pain point of agents that forget context or require manual retraining.
• Multi-Platform Messaging Gateway: A single gateway process connects Telegram, Discord, Slack, WhatsApp, Signal, and CLI with voice memo transcription and cross-platform conversation continuity. Eliminates the need to manage separate bots for each platform.
• Scheduled Automations: Built-in cron scheduler delivers daily reports, nightly backups, and weekly audits in natural language to any platform. Replaces fragile scripting with agent-driven, self-healing automation.
• Delegation & Parallelization: Spawn isolated subagents for parallel workstreams and write Python scripts that call tools via RPC to collapse multi-step pipelines. Solves the bottleneck of single-threaded agent execution.
• Seven Terminal Backends: Supports Local, Docker, SSH, Singularity, Modal, Daytona, and Vercel Sandbox with serverless persistence that hibernates when idle. USP: Deployable on anything from a $5 VPS to serverless infrastructure with near-zero idle cost.

Technical Architecture

Component	Details
Language	Python 3.11
Package Manager	uv (fast Python resolver)
Dependencies	Node.js, ripgrep, ffmpeg
LLM Providers	Nous Portal, OpenRouter (200+ models), NVIDIA NIM, Xiaomi MiMo, z.ai/GLM, Kimi/Moonshot, MiniMax, Hugging Face, OpenAI, custom endpoints
Memory & Search	FTS5 full-text search with LLM-powered summarization
User Modeling	Honcho dialectic system for persistent personality/behavior learning
Skill Standard	Compatible with agentskills.io open standard
Integration	MCP server support
Interface	Real terminal TUI with multiline editing, slash-command autocomplete, streaming tool output
Messaging	Single gateway process for Telegram, Discord, Slack, WhatsApp, Signal, CLI

Quick Start Guide

# Linux, macOS, WSL2, Termux installation
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

# Windows (native, PowerShell) installation
irm https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.ps1 | iex

# After installation
source ~/.bashrc    # reload shell (or: source ~/.zshrc)
hermes              # start chatting!

# Key commands
hermes model        # Choose your LLM provider and model
hermes tools        # Configure which tools are enabled
hermes config set   # Set individual config values
hermes gateway      # Start the messaging gateway
hermes setup        # Run the full setup wizard
hermes update       # Update to the latest version
hermes doctor       # Diagnose any issues

Pros, Cons & Use Cases

Pros

• Model-agnostic: No lock-in to any single provider; supports 200+ models through OpenRouter
• Minimal infrastructure: Runs on a $5 VPS with serverless persistence that hibernates when idle
• Research-ready: Built-in support for batch trajectory generation and RL environments
• Extensive platform support: Messaging gateway covers 6 platforms plus CLI
• Self-improving: Autonomous skill creation and refinement reduce manual intervention

Cons

• Windows support: Native Windows is early beta with rough edges; browser dashboard chat pane requires WSL2
• Android/Termux: Manual setup required, lacks full voice dependencies
• Complexity: The breadth of features (7 backends, multiple platforms, learning loops) may overwhelm casual users

Who should NOT use this?

• Users needing a simple, single-purpose chatbot – the learning loop and multi-platform complexity are overkill
• Teams requiring native Windows-only deployment without WSL2
• Users who prefer managed SaaS solutions and don't want to self-host infrastructure
• Projects with strict security policies against external LLM API calls or self-modifying code

Ideal Use Cases

• Personal automation assistant: Cross-platform messaging, scheduled reports, file management, and research that improves over time
• Research & RL environments: Batch trajectory generation, reinforcement learning experiments with agent skill evolution
• DevOps & infrastructure management: Deploy on a $5 VPS for monitoring, backups, and automated incident response
• Multi-platform customer support: Single agent backend serving Telegram, Discord, Slack, and WhatsApp simultaneously

Community & Activity

With 141,797 stars and active development through May 2026, Hermes Agent has clearly struck a chord with the developer community. This is not a dormant project – the latest update on 2026-05-10 signals ongoing refinement and feature development. The project's momentum reflects a strong demand for self-improving, deployable AI agents that break free from single-model lock-in. The combination of research-grade capabilities (RL environments, batch trajectory generation) with practical automation features (scheduled tasks, multi-platform messaging) has created a rare intersection that appeals to both researchers and practitioners.