From OpenClaw to Hermes | 从＂AI工具＂到＂AI同事＂

# apt-get install git

$ brew install git

$ sudo port install git

wsl --install

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

wayne@WayneYang:~$ curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash┌─────────────────────────────────────────────────────────┐│             ⚕ Hermes Agent Installer                    │├─────────────────────────────────────────────────────────┤│  An open source AI agent by Nous Research.              │└─────────────────────────────────────────────────────────┘✓ Detected: linux (ubuntu)→ Checking for uv package manager...✓ uv found (uv 0.11.7 (x86_64-unknown-linux-gnu))→ Checking Python 3.11...✓ Python found: Python 3.11.15→ Checking Git...✓ Git 2.43.0 found→ Checking Node.js (for browser tools)...✓ Node.js v22.22.2 found→ Checking ripgrep (fast file search)...→ Checking ffmpeg (TTS voice messages)...→ sudo is needed ONLY to install optional system packages (ripgrep ffmpeg) via your package manager.→ Hermes Agent itself does not require or retain root access.Install ripgrep for faster file search ffmpeg for TTS voice messages? [Y/n] Y[sudo] password for wayne:Reading package lists... DoneBuilding dependency tree... DoneReading state information... DoneE: Unable to locate package ripgrepE: Unable to locate package ffmpeg⚠ ripgrep not installed (file search will use grep fallback)→ To install ripgrep manually:→   sudo apt install ripgrep⚠ ffmpeg not installed (TTS voice messages will be limited)→ To install ffmpeg manually:→   sudo apt install ffmpeg→ Installing to /home/wayne/.hermes/hermes-agent...→ Trying SSH clone...→ SSH failed, trying HTTPS...Cloning into '/home/wayne/.hermes/hermes-agent'...fatal: unable to access 'https://github.com/NousResearch/hermes-agent.git/': GnuTLS recv error (-110): The TLS connection was non-properly terminated.✗ Failed to clone repositorywayne@WayneYang:~$ curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash┌─────────────────────────────────────────────────────────┐│             ⚕ Hermes Agent Installer                    │├─────────────────────────────────────────────────────────┤│  An open source AI agent by Nous Research.              │└─────────────────────────────────────────────────────────┘✓ Detected: linux (ubuntu)→ Checking for uv package manager...✓ uv found (uv 0.11.7 (x86_64-unknown-linux-gnu))→ Checking Python 3.11...✓ Python found: Python 3.11.15→ Checking Git...✓ Git 2.43.0 found→ Checking Node.js (for browser tools)...✓ Node.js v22.22.2 found→ Checking ripgrep (fast file search)...→ Checking ffmpeg (TTS voice messages)...→ Installing ripgrep ffmpeg...Reading package lists... DoneBuilding dependency tree... DoneReading state information... DoneE: Unable to locate package ripgrepE: Unable to locate package ffmpeg⚠ ripgrep not installed (file search will use grep fallback)→ To install ripgrep manually:→   sudo apt install ripgrep⚠ ffmpeg not installed (TTS voice messages will be limited)→ To install ffmpeg manually:→   sudo apt install ffmpeg→ Installing to /home/wayne/.hermes/hermes-agent...→ Trying SSH clone...→ SSH failed, trying HTTPS...Cloning into '/home/wayne/.hermes/hermes-agent'...remote: Enumerating objects: 49216, done.remote: Counting objects: 100% (561/561), done.remote: Compressing objects: 100% (305/305), done.remote: Total 49216 (delta 406), reused 256 (delta 256), pack-reused 48655 (from 5)Receiving objects: 100% (49216/49216), 135.55 MiB | 6.74 MiB/s, done.Resolving deltas: 100% (34698/34698), done.✓ Cloned via HTTPS✓ Repository ready→ Creating virtual environment with Python 3.11...Using CPython 3.11.15Creating virtual environment at: venvActivate with: source venv/bin/activate✓ Virtual environment ready (Python 3.11)→ Installing dependencies...→ Some build tools may be needed for Python packages...✓ Build tools installed✓ Main package installed✓ All dependencies installed→ Installing Node.js dependencies (browser tools)...

source ~/.bashrc   # or: source ~/.zshrchermes             # Start chatting!

hermes model          # Choose your LLM provider and modelhermes tools          # Configure which tools are enabledhermes gateway setup  # Set up messaging platformshermes config set# Set individual config valueshermes setup          # Or run the full setup wizard to configure everything a

和大多数 Agent 最大的不同在于，Hermes 并不是“做完就结束”，而是围绕一个持续循环在运转——做 → 学 → 再改进。

首先，是它的自我进化能力。

每一次任务完成后，Hermes 不只是给出结果，还会拆解哪些步骤是有效的，并自动沉淀为结构化的 Skill 文档，在后续任务中直接复用。背后依托的是 Nous Research 的强化学习框架 Atropos，这意味着系统不仅在用模型，还在不断“训练自己如何更好地使用模型”。

其次，是可持续的记忆系统。

不同于一次性对话，Hermes 会将历史记录存储下来，并通过检索与摘要机制，在后续任务中调用相关信息。它的记忆是分层的——从即时对话，到长期笔记，再到用户习惯与技能沉淀，让 Agent 真正具备“跨任务连续工作”的能力。

再往上一层，是它的技能体系。

在 Hermes 中，能力不是写死的，而是以“技能”的形式按需加载。系统可以自动生成技能，也支持人工补充，并在实际使用中不断优化。这种方式既保证了灵活性，也降低了推理成本。

在工程层面，Hermes 也做了更完整的设计。

它支持多平台接入（如即时通讯工具）、多环境运行（本地、容器、无服务器等），可以根据使用场景灵活部署。同时，在安全性上，从权限控制、命令审批到运行环境隔离，都做了系统级约束，而不是事后补丁。

最后，是它的开放性。

Hermes 完全开源，支持接入多种模型与工具，开发者可以根据需求自由扩展，而不是被锁定在单一生态中。