mcpField
K

Keyban Agent Wallet

本地服务
keyban.io金融数据

关于 Keyban Agent Wallet

为 AI Agent 配置其专属区块链钱包:查询余额、USDC 转账,以及在策略约束下的 x402 支付。

配置

本地服务的配置在各客户端中通用,复制后粘贴到客户端的 MCP 配置文件即可。

mcp.json · 通用配置
{
"mcpServers": {
"io-keyban-mcp-agent-wallet": {
"command": "npx",
"args": [
"-y",
"@keyban/cli"
],
"env": {
"KEYBAN_API_KEY": "<KEYBAN_API_KEY>",
"KEYBAN_APP_ID": "<KEYBAN_APP_ID>",
"KEYBAN_INVITE_KEY": "<KEYBAN_INVITE_KEY>",
"KEYBAN_WALLET_ADDRESS": "<KEYBAN_WALLET_ADDRESS>",
"KEYBAN_FUNDING": "<KEYBAN_FUNDING>",
"KEYBAN_NETWORK": "<KEYBAN_NETWORK>",
"KEYBAN_BASE_URL": "<KEYBAN_BASE_URL>",
"KEYBAN_DKG_BLOB": "<KEYBAN_DKG_BLOB>",
"MCP_LOG_FILE": "<MCP_LOG_FILE>",
"MCP_LOG_LEVEL": "<MCP_LOG_LEVEL>"
}
}
}
}
启动命令npx -y @keyban/cli
需要配置的环境变量
KEYBAN_API_KEY密钥Keyban API key of the application the wallet belongs to. Never exposed to the model.
KEYBAN_APP_ID密钥Identifier of the Keyban application.
KEYBAN_INVITE_KEY密钥Wallet mode only. One-shot ECDSA private key (0x + 64 hex) authorizing the first-boot registration of this client as a FROST signer of the wallet.
KEYBAN_WALLET_ADDRESS密钥Wallet mode only. Address of the deployed SchnorrAccount this server drives (0x + 40 hex).
KEYBAN_FUNDING密钥Who pays for the payments: wallet (the agent's own wallet) or prepaid (the organization's prepaid credit, with no wallet address or invite key).
KEYBAN_NETWORK密钥Blockchain network the wallet operates on, for example base or xdc.
KEYBAN_BASE_URL密钥Base URL of the Keyban backend. Accepts a full URL, or one of the environment aliases dev, testing, staging, prod and sandbox.
KEYBAN_DKG_BLOB密钥Optional pre-generated FROST client share, base64-encoded JSON. Used by hosted runtimes with no OS keyring; it does not replace the invite key.
MCP_LOG_FILEPath the server writes its log to.
MCP_LOG_LEVELLog verbosity.

本地服务需要在你的电脑上安装运行环境(如 Node.js 18+)。各客户端的配置文件位置不同,请参考客户端文档。

安装包
npm@keyban/cli0.2.1

工具

2

收录的工具列表 · 更新于 1 天前

此工具无需参数。

详细介绍

维护者还没有提供详细介绍,可以先查看代码仓库中的 README。

评论

支持 Markdown

第一次接触 MCP?

从协议是什么,到在客户端里装上第一个服务,这几条回答覆盖了大多数入门问题。

一个开放协议,规定了 AI 应用如何以统一方式连接外部工具与数据源。客户端负责与模型对话,MCP 服务负责提供工具、资源和提示模板,两者通过标准消息通信,同一个服务可以被不同客户端复用。