Zerion Hosted MCP
Zerion MCP let's your agent analyze EVM and Solana wallets and summarize total portfolio value, top holdings, DeFi positions, recent transactions, PnL and more.
README
zerion-ai
Maintained by Zerion.
zerion-ai is the public, self-contained repo for using Zerion from AI agents and developer tools.
It packages two first-class integration paths:
- Hosted MCP for Cursor, Claude, and other MCP-native agent environments
zerion-clifor OpenClaw-like and command-based agent runtimes
It also ships one flagship workflow:
wallet-analysisas a reusable skill/playbook for portfolio, positions, transactions, and PnL analysis
1. Choose your authentication method
Option A: API Key
Get an API key and export it: Get your API key
export ZERION_API_KEY="zk_dev_..."
- API auth via HTTP Basic Auth
- dev keys beginning with
zk_dev_ - current dev-key limits of 120 requests/minute and 5k requests/day
Useful docs:
Option B: x402 Pay-per-call
No API key needed. Pay $0.01 USDC per request on Base via the x402 protocol. The CLI handles the payment handshake automatically using your wallet's private key.
Setup:
export WALLET_PRIVATE_KEY="0x..." # EVM wallet with USDC on Base
Then use the --x402 flag:
zerion-cli wallet analyze 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045 --x402
Or enable x402 globally:
export ZERION_X402=true
zerion-cli wallet analyze 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045
2. Install skills (Claude Code, Cursor, OpenClaw)
npx skills add zeriontech/zerion-ai
This installs 3 skills into your agent:
| Skill | Description |
|---|---|
| wallet-analysis | Analyze wallets: portfolio, positions, transactions, PnL |
| chains | List supported blockchain networks |
| zerion-cli | CLI setup, authentication, and troubleshooting |
The skills reference zerion-cli which runs via npx zerion-cli (no global install needed).
3. Choose your integration path
MCP clients
Use this if your agent runtime already supports MCP.
Start here:
OpenClaw and CLI-based agents
Use this if your framework models tools as shell commands returning JSON.
npm install -g zerion-cli
zerion-cli wallet analyze 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045
Start here:
4. Run the first wallet analysis
MCP quickstart
-
Export your API key:
export ZERION_API_KEY="zk_dev_..." -
Add the hosted Zerion MCP config from examples/cursor/mcp.json or examples/claude/mcp.json
-
Ask:
Analyze the wallet 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045. Summarize total portfolio value, top positions, recent transactions, and PnL.
CLI quickstart
With API key:
npm install -g zerion-cli
export ZERION_API_KEY="zk_dev_..."
zerion-cli wallet analyze 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045
With x402 (no API key needed):
npm install -g zerion-cli
export WALLET_PRIVATE_KEY="0x..."
zerion-cli wallet analyze 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045 --x402
Example output:
{
"wallet": {
"query": "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045"
},
"portfolio": {
"total": 450000,
"currency": "usd"
},
"positions": {
"count": 42
},
"transactions": {
"sampled": 10
},
"pnl": {
"available": true
}
}
Example wallets
This repo uses the same public wallets across examples:
vitalik.eth/0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045- ENS DAO treasury /
0xFe89Cc7Abb2C4183683Ab71653c4cCd1b9cC194e - Aave collector /
0x25F2226B597E8F9514B3F68F00F494CF4F286491
What ships in this repo
skills/: 3 agent skills installable vianpx skills add zeriontech/zerion-aiwallet-analysis/: portfolio, positions, transactions, and PnL analysischains/: supported blockchain networks referencezerion-cli/: CLI setup, auth, and troubleshooting
mcp/: hosted Zerion MCP setup plus the tool catalogcli/:zerion-cliJSON-first CLI (published to npm)examples/: Cursor, Claude, OpenAI Agents SDK, raw HTTP, and OpenClaw setups
Failure modes to expect
Both the MCP and CLI surfaces should handle:
- missing or invalid API key
- invalid wallet address
- unsupported chain filter
- empty wallets / no positions
- rate limits (
429) - upstream timeout or temporary unavailability
See mcp/README.md and cli/README.md for the concrete behavior used in this repo.
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。