hedera-mcp-server
Integrates with a Langchain wrapper to interact with the Hedera network via natural language commands.
README
Hedera MCP Server
This project provides a server that integrates with a Langchain wrapper to interact with the Hedera network. It allows users to perform Hedera operations through natural language commands facilitated by the Langchain setup.
⚠️ Attention
Since the current version on the master branch was moved from another repository, it temporarily includes the hedera-agent-kit package, which hasn't been officially released yet. Therefore, it is included as a dependency pulled from GitHub.
Prerequisites
Before you begin, ensure you have the following installed:
- Node.js (20 or later recommended)
- pnpm (
npm install -g pnpm) - Git
Quickstart
The whole thing consists of two services that we need to run.
-
Clone the repository:
git clone https://github.com/mateuszm-arianelabs/hedera-mcp-server.git cd hedera-mcp-server -
Set up environment variables in packages:
- You need to copy the .env.example file and create .env from it in two locations:
- packages/langchain-proxy
- packages/mcp-server
- You need to copy the .env.example file and create .env from it in two locations:
Edit the .env file and fill in the required configuration values (e.g., Hedera keys, network details).
API_URL= # url to langchain proxy endpoint
PORT= # port on which MCP server will be started
MCP_AUTH_TOKEN= # array of accepted tokens separated by comas
LANGCHAIN_PROXY_TOKEN= # token for accessing the Langchain proxy
-
Install dependencies:
pnpm install -
Start Lang chain Proxy Service
pnpm run dev:lc(Or
pnpm startfor production mode, if configured). -
Run the mcp server:
pnpm run dev:mcp(Or
pnpm startfor production mode, if configured).
Available Tools
- Interact with Hedera: This core functionality enables the server to make calls to the running Langchain wrapper service. The wrapper then interprets these requests and performs the corresponding actions on the Hedera network (e.g., creating tokens, transferring HBAR, interacting with smart contracts).
Configuration for Visual Studio Code
To establish a connection with an MCP server, the X-MCP-AUTH-TOKEN header is required.
Below is an example .vscode/settings.json configuration for Visual Studio Code:
{
"mcp": {
"inputs": [],
"servers": {
"hedera": {
"type": "sse",
"url": "http://localhost:3000/sse",
"headers": { "X-MCP-AUTH-TOKEN": "your-mcp-auth-token"}
}
}
}
}
Note: Currently, passing additional headers (such as X-MCP-AUTH-TOKEN) is not supported in Cursor IDE.
Source: Cursor Forum
However, MCP integration is evolving rapidly, and support for custom headers is expected to be added in future versions of Cursor and other MCP Client Tools.
Potential Transition to OAuth
As of March 2025, the official MCP OAuth specification has been published, but it still has several limitations and implementation challenges. A revision of the approach is likely (source).
In preparation for future OAuth integration, here are some example repositories that showcase MCP + OAuth setups:
These can be used as references when considering OAuth-based authentication for MCP servers in future development phases.
Simplified Architecture

API Documentation (langchain-proxy)
GET /sse
Establishes a Server-Sent Events (SSE) connection with the MCP server.
Request
GET /sse
X-MCP-AUTH-TOKEN: example-token
| Header | Type | Description |
|---|---|---|
| X-MCP-AUTH-TOKEN | string | Required authentication token that must match one of the tokens in servers env file (MCP_AUTH_TOKEN variable) |
Response
On successful authentication, establishes an SSE connection with content type text/event-stream.
Status Codes
200 OK: Connection established successfully
401 Unauthorized: Invalid or missing authentication token
Error response
{
"content": [
{
"type": "text",
"content": "Unauthorized: Invalid or missing X-MCP-AUTH-TOKEN header"
}
]
}
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。