Acornio MCP Server
A Model Context Protocol server that provides access to knowledge base documents from GitHub repositories.
README
Acornio MCP Server
A Model Context Protocol (MCP) server that provides access to knowledge base documents from GitHub repositories.
Features
- 4 Tools:
listKnowledgeBases,listDocuments,searchDocuments,getDocument - Multiple Knowledge Bases: Supports multiple GitHub repositories as knowledge sources
- Auto-discovery: Automatically discovers markdown documents in configured repositories
- Stateless: Works on any hosting platform (Railway, Vercel, etc.)
Deployment
This is a Next.js application that can be deployed to any platform that supports Next.js.
Quick Deploy Options
- Vercel: Connect your GitHub repo and deploy with one click
- Railway: Connect your repo and Railway auto-detects Next.js
- Netlify: Connect your repo and configure build command:
npm run build - Fly.io: Use the Next.js template
- Any Node.js host: Run
npm run build && npm start
Deployment Steps
- Push your code to GitHub
- Connect your repository to your chosen hosting platform
- Add environment variables:
GITHUB_TOKEN(optional, for higher API rate limits)MCP_AUTH_TOKEN(optional, for securing access to your MCP server)
- Deploy! The platform should auto-detect Next.js and configure accordingly
Platform-Specific Notes
- Vercel: Zero configuration needed, just connect and deploy
- Railway: Auto-detects Next.js, no additional config required
- Netlify: May need to set build command:
npm run buildand publish directory:.next - Self-hosted: Ensure Node.js 18+ is installed and set
NODE_ENV=production
Configuration
Edit app/api/[transport]/route.ts to configure your knowledge bases:
const KNOWLEDGE_BASES = [
{
id: 'vision',
name: 'My Knowledge Base',
description: 'Description of knowledge base',
owner: 'github-username',
repo: 'repo-name',
branch: 'main',
path: './docs', // path to markdown files
},
// Add more...
];
Security & Authentication
This MCP server supports optional token-based authentication to secure access to your knowledge bases.
How It Works
- When
MCP_AUTH_TOKENis not set, the server accepts all requests (useful for local development or private networks) - When
MCP_AUTH_TOKENis set, clients must provide this token in theAuthorizationheader - The server accepts both
Bearer <token>and<token>formats
Setting Up Authentication
- Generate a secure random token (e.g.,
openssl rand -hex 32) - Set the
MCP_AUTH_TOKENenvironment variable on your hosting platform - Configure your MCP client with the same token (see examples below)
Best Practices
- Use a strong, randomly-generated token (at least 32 characters)
- Keep your token secret and never commit it to version control
- Rotate your token periodically
- Use HTTPS for all production deployments to protect the token in transit
Environment Variables
| Variable | Description | Required |
|---|---|---|
GITHUB_TOKEN |
GitHub personal access token for higher API rate limits (60/hr → 5000/hr) | No |
MCP_AUTH_TOKEN |
Authentication token for securing MCP server access. When set, clients must provide this token in the Authorization header. | No |
Connect to MCP Client
Cursor / Claude Desktop
Add to your MCP config (replace https://your-app.example.com with your deployed URL):
{
"mcpServers": {
"acornio-knowledge": {
"url": "https://your-app.example.com/api/mcp"
}
}
}
If you've configured MCP_AUTH_TOKEN, add authentication:
{
"mcpServers": {
"acornio-knowledge": {
"url": "https://your-app.example.com/api/mcp",
"headers": {
"Authorization": "Bearer your_token_here"
}
}
}
}
Or use mcp-remote for stdio transport:
{
"mcpServers": {
"acornio-knowledge": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://your-app.example.com/api/mcp"]
}
}
}
With authentication:
{
"mcpServers": {
"acornio-knowledge": {
"command": "npx",
"args": ["-y", "mcp-remote", "--header", "Authorization: Bearer your_token_here", "https://your-app.example.com/api/mcp"]
}
}
}
Note: Replace your-app.example.com with your actual deployment URL:
- Vercel:
your-app.vercel.app - Railway:
your-app.railway.app - Netlify:
your-app.netlify.app - Custom domain:
your-domain.com
Local Development
npm install
npm run dev
Server runs at http://localhost:3000/api/mcp
Available Tools
listKnowledgeBases
Lists all configured knowledge bases and their document counts.
listDocuments
Lists all documents, optionally filtered by knowledge base.
searchDocuments
Searches for documents by name/title across knowledge bases.
getDocument
Retrieves the full content of a specific document.
License
MIT
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。