Cloud · MODEL CONTEXT PROTOCOL
lmcmz flow mcp server
Model Context Protocol (MCP) server for Flow blockchain with direct RPC communication
连接能力
lmcmz flow mcp server 为 AI Agent 提供什么
- Balance checking for Flow and fungible tokens
- Domain resolution for `.find` and `.fn` domains
- Script execution for reading blockchain data
- Transaction submission and monitoring
- Account information retrieval
- Flow configuration based on environment variables
- `PORT` - HTTP port to listen on (if not set, defaults to stdio mode)
- `FLOW_NETWORK` - Flow network to connect to (mainnet, testnet, emulator)
- `FLOW_ACCESS_NODE` - Custom Flow access node URL
- `LOG_LEVEL` - Logging level (debug, info, warn, error)
- `/sse` - Server-Sent Events endpoint for real-time updates
- `/messages` - POST endpoint for sending tool requests
- `/health` - Health check endpoint
- `/` - Server information
- `get_balance` - Get Flow balance for an address
- `get_token_balance` - Get fungible token balance
- `get_account` - Get account information
- `resolve_domain` - Resolve a .find or .fn domain to an address
- `execute_script` - Execute a Cadence script
- `send_transaction` - Send a transaction to the blockchain
- `get_transaction` - Get transaction details by ID
- **v0.1.1** - Bug fix for formatArguments import issue in transaction service
- **v0.1.0** - Initial release with basic Flow blockchain integration
安全
MCP 收录不等于安全审核
MCP 服务器可能获得模型上下文、凭据、本地文件或调用外部系统的权限。连接 Agent 前,请检查代码、环境变量、网络行为、软件包来源与维护状态。
README
仓库文档
A Model Context Protocol (MCP) server for interacting with the Flow blockchain. This server enables AI assistants to access Flow blockchain data and perform operations through a standardized interface.
- Balance checking for Flow and fungible tokens
- Domain resolution for
.findand.fndomains - Script execution for reading blockchain data
- Transaction submission and monitoring
- Account information retrieval
- Flow configuration based on environment variables
npx -y @outblock/flow-mcp-server --stdioOr specify HTTP mode with a port:
npx -y @outblock/flow-mcp-server --port 3000git clone https://github.com/lmcmz/flow-mcp-server.git
cd flow-mcp-server
npm install
npm run build
npm start# Run in stdio mode (for AI assistant integration)
npx -y @outblock/flow-mcp-server --stdio
# Run as HTTP server on port 3000
npx -y @outblock/flow-mcp-server --port 3000
# Specify Flow network
npx -y @outblock/flow-mcp-server --port 3000 --network testnet# Run in stdio mode
npm run build
npm start
# Run as HTTP server on port 3000
npm run build
PORT=3000 npm start
# For development with auto-reload
npm run devThe server can be configured using environment variables:
PORT- HTTP port to listen on (if not set, defaults to stdio mode)FLOW_NETWORK- Flow network to connect to (mainnet, testnet, emulator)FLOW_ACCESS_NODE- Custom Flow access node URLLOG_LEVEL- Logging level (debug, info, warn, error)
When integrating with AI assistants like Claude, you can start the MCP server in stdio mode and connect it to your assistant's tool configuration.
Example Claude tool configuration:
{
"tools": [
{
"name": "flow-mcp-server",
"command": "npx -y @outblock/flow-mcp-server --stdio"
}
]
}When running in HTTP mode, the following endpoints are available:
/sse- Server-Sent Events endpoint for real-time updates/messages- POST endpoint for sending tool requests/health- Health check endpoint/- Server information
Example HTTP API call:
curl -X POST http://localhost:3000/messages \
-H "Content-Type: application/json" \
-d '{
"tool_request": {
"name": "get_balance",
"parameters": {
"address": "0x2d4c3caffbeab845",
"network": "mainnet"
}
}
}'get_balance- Get Flow balance for an addressget_token_balance- Get fungible token balanceget_account- Get account informationresolve_domain- Resolve a .find or .fn domain to an addressexecute_script- Execute a Cadence scriptsend_transaction- Send a transaction to the blockchainget_transaction- Get transaction details by ID
- v0.1.1 - Bug fix for formatArguments import issue in transaction service
- v0.1.0 - Initial release with basic Flow blockchain integration
MIT
社区