automation

Hermes One

Open-source desktop AI agent that writes its own skills, remembers across sessions, and runs scheduled tasks on any platform.

Try Hermes One →Visit website
7.4/10

How the 7.4 was reached

  1. Quality of output7/10

    14,324 stars, 1,617 forks, active MIT project; community-maintained with 415 open issues. github.com

  2. Ease of use8/10

    Guided first-run setup with no terminal needed, on all major platforms. hermesone.org

  3. Pricing value10/10

    Free, MIT licensed, though model provider usage may cost extra. github.com

  4. API and integration quality6/10

    Many model providers, 14 toolsets and messaging platforms; no public API documented. hermesone.org

  5. Solves the problem it claims to solve6/10

    14k stars show adoption as a desktop front end; the feature claims are vendor-only. github.com

7.4/10 is the mean of the 5 criteria that apply. Scored on the five-criteria rubric v1, 29 September 2026. How scoring works

Pricing

Free
$0
  • Full feature access
  • Open source
  • Self-hosted
  • All 14 toolsets
  • Unlimited model connections

Key Features

  • Closed learning loop: agent creates and refines skills automatically after complex tasks
  • Multi-platform messaging gateway for Telegram, Discord, Slack, WhatsApp, Signal, and CLI
  • Built-in cron scheduler for unattended daily reports, backups, and audits
  • Seven terminal backends including local, Docker, SSH, Modal, and Vercel Sandbox
  • Model-agnostic: connects to OpenRouter (300+ models), OpenAI, Anthropic, Gemini, Ollama, and custom endpoints
  • FTS5 session search with LLM summarization for cross-session memory recall
  • Subagent delegation for parallel workstreams via Python RPC scripting

Pros & Cons

Pros

  • Completely free and open source with 14k+ GitHub stars backing active development
  • The skill-creation learning loop is a real differentiator — most agents don't persist and improve across sessions
  • Model-agnostic design means no vendor lock-in; switch providers with one command
  • Multi-platform delivery from a single gateway covers messaging apps most agents ignore

Cons

  • Technical setup barrier: CLI origins mean non-developers will struggle with initial configuration
  • Self-improving skill system is opaque — hard to audit what the agent has learned or unlearn bad habits
  • No managed hosting; you supply your own model API keys and infrastructure
  • Community-maintained desktop wrapper means the core Nous Research agent and the GUI can drift out of sync
Verdict

Hermes One is the most serious open-source attempt at a persistent, self-improving desktop agent. The learning loop and cross-platform messaging set it apart from chat wrappers, but the payoff requires genuine technical investment to configure and trust. A strong pick for developers who want an agent that compounds knowledge over time; not suitable for users who want something that works out of the box.

Try Hermes One →

Added to scored.tools on

Competitors to Hermes One

Other tools in the automation category worth comparing.

Jan

8.6/10
coding

Open-source ChatGPT alternative that runs AI models locally or connects to cloud services

Open WebUI

8.6/10
coding

Self-hosted AI platform that connects any model with full data control and Python extensibility.

AnythingLLM

8.2/10
productivity

All-in-one AI application for chatting with documents, AI agents, and custom models - fully private and local.

AutoGPT

6.6/10
automation

Create intelligent AI assistants that streamline your digital workflow and automate repetitive tasks.

Cherry Studio

7.6/10
coding

Open-source multi-model AI desktop client supporting 300+ AI models across Windows, macOS, and Linux

More Articles Featuring Hermes One

review

Hermes One Review 2026: An Honest Look at the Self-Improving Desktop Agent

Hermes One is a free, open-source desktop agent that writes its own skills and remembers across sessions. Here's what works, what doesn't, and who should run it.

Stay sharp on AI tools

Weekly picks, new reviews, and deals. No spam.