keygenix-mcp
Non-custodial TEE key management and signing for AI agents, supporting multiple blockchains.
README
keygenix-mcp
Keygenix MCP Server — Non-custodial TEE key management & signing for AI agents.
Private keys are generated, stored, and used exclusively inside a Trusted Execution Environment (TEE). They never leave in plaintext — not to you, not to Keygenix, not to the AI.
Quick Start
1. Get your credentials
- Register at keygenix.pro
- Create an organization → note
orgCode - Create a wallet → note
walletCode - Generate two keypairs (run once):
# Use the CLI to generate keypairs (easiest)
git clone https://github.com/wallet-io/keygenix-skill
cd keygenix-skill/cli && npm install
node client.js keygen # → copy publicKey as API Auth Key
node client.js keygen # → copy publicKey as AuthKey (separate keypair)
- Register both public keys in the Keygenix dashboard.
2. Configure your AI client
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"keygenix": {
"command": "node",
"args": ["/path/to/node_modules/keygenix-mcp/dist/index.js"],
"env": {
"KEYGENIX_API_PRIV_KEY": "your-api-auth-private-key-hex",
"KEYGENIX_AUTH_PRIV_KEY": "your-authkey-private-key-hex",
"KEYGENIX_ORG_CODE": "your-org-code",
"KEYGENIX_WALLET_CODE": "your-wallet-code"
}
}
}
}
Once published to npm, replace with
"command": "npx", "args": ["keygenix-mcp"]
Cursor / Windsurf
Edit .cursor/mcp.json or .windsurf/mcp.json:
{
"mcpServers": {
"keygenix": {
"command": "node",
"args": ["/path/to/node_modules/keygenix-mcp/dist/index.js"],
"env": {
"KEYGENIX_API_PRIV_KEY": "...",
"KEYGENIX_AUTH_PRIV_KEY": "...",
"KEYGENIX_ORG_CODE": "...",
"KEYGENIX_WALLET_CODE": "..."
}
}
}
}
OpenClaw
Add to your OpenClaw MCP config, or use the keygenix OpenClaw Skill directly.
Available Tools
| Tool | Description |
|---|---|
keygen |
Generate a new secp256k1 keypair locally (no network) |
list_keys |
List all keys in the wallet |
get_key |
Get details of a key by keyCode |
create_key |
Create a new key (mnemonic/private/secret) |
import_key |
Import existing key into TEE (ECIES encrypted) |
list_addresses |
List derived addresses for a key |
create_address |
Derive a new address for a chain |
sign_transaction |
Sign a blockchain transaction (EVM/SOL/SUI/etc.) |
sign_message |
Sign an arbitrary message |
Supported Chains
EVM · Solana · Bitcoin · Litecoin · Dogecoin · Zcash · Tron · Ripple · Sui · TON · Cardano · Aptos · Cosmos · Sei
Security Model
AI Agent
↓ calls MCP tool (no keys in prompt)
keygenix-mcp (local process)
↓ ECDSA-signed HTTPS requests
Keygenix TEE API
↓ private key never leaves enclave
Signed transaction returned
- API Auth Private Key — signs every API request. Store in env, never hardcode.
- AuthKey Private Key — authorizes sign/export. Signed locally; Keygenix only sees the public key.
- Private keys — generated inside TEE, never exposed in plaintext.
Development
git clone https://github.com/wallet-io/keygenix-mcp
cd keygenix-mcp
npm install
npm run build
npm start
Distribution
| Channel | Command |
|---|---|
| GitHub | npm install github:wallet-io/keygenix-mcp |
| npm (coming soon) | npx keygenix-mcp |
| OpenClaw | clawhub install keygenix |
| Smithery | smithery.ai/server/keygenix |
Links
- Website: keygenix.pro
- API Docs: keygenix.pro/docs.html
- Skill + CLI: wallet-io/keygenix-skill
- Issues: github.com/wallet-io/keygenix-mcp/issues
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。