hermes-mcp-server
Bridges MCP-compatible AI clients to over 100 Hermes Agent skills for web browsing, file management, code execution, GitHub operations, and more.
README
🔌 Hermes MCP Server
Expose 100+ Hermes Agent skills to any MCP-compatible AI client.
Connect Claude Desktop, Cursor, Continue, Zed, or any MCP client to Hermes Agent's powerful toolkit.
✨ What You Get
Install one MCP server, get ALL these tools in your AI client:
| Category | Tools |
|---|---|
| 🌐 Web | Search, extract, navigate, scrape |
| 🖥️ Browser | Full headless browser — click, type, scroll, screenshot |
| 💻 Terminal | Execute shell commands, run Python code |
| 📁 Files | Read, write, search files (ripgrep-backed) |
| 🐙 GitHub | Repo info, search, create issues, manage PRs |
| 📚 Research | arXiv search, YouTube transcripts, blog monitoring |
| 📝 Productivity | Notion, Google Workspace, Linear, Airtable |
| 🎨 Creative | Image generation, ASCII art, diagrams, music |
| 🤖 AI | Full Hermes Agent reasoning — multi-step, tool-using |
🚀 Quick Start
1. Install
pip install hermes-mcp-server
2. Start Hermes Agent
# Start Hermes Agent with HTTP API enabled
hermes serve --http --port 8080
3. Configure your MCP client
Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"hermes": {
"command": "hermes-mcp",
"args": ["--endpoint", "http://localhost:8080"],
"env": {
"HERMES_API_KEY": "your-api-key-here"
}
}
}
}
Cursor / Continue / Zed:
{
"mcpServers": {
"hermes": {
"command": "hermes-mcp",
"args": ["--endpoint", "https://your-hermes-instance.com"]
}
}
}
4. Restart your MCP client and start using Hermes tools!
📋 Complete Tool List
| Tool | Description |
|---|---|
web_search |
Google/Bing web search with snippets |
web_extract |
Extract clean content from any URL |
browser_navigate |
Headless browser navigation |
browser_snapshot |
Get page accessibility tree |
browser_click |
Click elements by ref ID |
terminal_exec |
Execute shell commands |
code_execute |
Run Python code with Hermes SDK |
file_read |
Read files with line numbers |
file_write |
Create/overwrite files |
file_search |
Ripgrep-powered file/content search |
github_repo_info |
Get GitHub repo details |
github_search_repos |
Search GitHub repos |
github_create_issue |
Create GitHub issues |
arxiv_search |
Search academic papers |
youtube_transcript |
Get YouTube transcripts |
notion_search |
Search Notion workspace |
image_generate |
AI image generation |
ai_complete |
Full Hermes Agent reasoning |
80+ more tools coming soon — all 100+ Hermes skills will be exposed.
🐳 Docker
docker run -e HERMES_ENDPOINT=http://host.docker.internal:8080 \
ghcr.io/pangtongya/hermes-mcp-server:latest
💰 Monetization
Hermes MCP Server is free and open-source. Revenue comes from:
| Tier | Price | Includes |
|---|---|---|
| 🆓 Community | Free | Self-hosted, all tools, community support |
| ⭐ Pro | $5/mo | Priority support, hosted Hermes endpoint, 1000 req/day |
| 🏢 Team | $29/mo | 5 seats, hosted endpoint, 10000 req/day, Slack support |
| 🚀 Enterprise | Custom | Dedicated instance, SSO, SLA, unlimited requests |
🏗️ Architecture
┌──────────────────┐ MCP Protocol ┌──────────────────┐ HTTP ┌──────────────────┐
│ Claude Desktop │◄────────────────────►│ Hermes MCP │◄────────────►│ Hermes Agent │
│ Cursor / Zed │ (stdio / JSON-RPC) │ Server │ (REST API) │ 100+ Skills │
│ Continue / etc │ │ (Python) │ │ Tools + Models │
└──────────────────┘ └──────────────────┘ └──────────────────┘
🔧 Configuration
| Environment Variable | Default | Description |
|---|---|---|
HERMES_ENDPOINT |
http://localhost:8080 |
Hermes Agent HTTP endpoint |
HERMES_API_KEY |
(none) | API key for authentication |
🛠️ Development
git clone https://github.com/pangtongya/hermes-mcp-server.git
cd hermes-mcp-server
pip install -e ".[dev]"
# Run tests
pytest
# Run locally
python -m hermes_mcp.cli --endpoint http://localhost:8080
📄 License
MIT © pangtongya
⭐ Star History
If this project helps you, please consider sponsoring or giving it a ⭐!
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。