agent-wallet-mcp
Enables agents to maintain sandboxed spend ledgers, enforce monthly budgets, transfer funds between agents, and generate invoices, all without real financial accounts.
README
<p align="center"> <img src="assets/logo.png" width="80" alt="AgentPay"> </p>
AgentWallet MCP
Your agent can maintain its own sandboxed spend ledger, enforce monthly budgets on itself, transfer funds to other agents, and generate formal invoices — all without touching your financial accounts.
The wallet is a local numeric ledger (stored in ~/.agentwallet/). It is not connected to any bank, payment processor, or blockchain. Use it to model agent-to-agent accounting and spending limits within your own infrastructure.
What your agent can do
- Create a wallet for any agent ID and fund it with an initial balance
- Check current balance, monthly spend total, and pending transactions at any time
- Deposit and withdraw from any wallet, with automatic budget enforcement on withdrawal
- Transfer between two agent wallets in a single atomic operation — both sides recorded
- Set a hard monthly spending cap: withdrawals and transfers that would exceed it are rejected at the time of the call
- Generate numbered invoice records (
INV-000001, etc.) for agent-to-agent service billing
Installation
Requires: Python 3.10+, mcp package.
pip install mcp
Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"agent-wallet": {
"command": "python",
"args": ["/absolute/path/to/agent-wallet-mcp/server.py"]
}
}
}
Cursor — add to .cursor/mcp.json in your project root:
{
"mcpServers": {
"agent-wallet": {
"command": "python",
"args": ["/absolute/path/to/agent-wallet-mcp/server.py"]
}
}
}
Tool Reference
| Tool | Description | Key params |
|---|---|---|
wallet_create |
Create a new wallet for an agent | agent_id (required), initial_balance (default: 0) |
wallet_balance |
Get balance, total spent, monthly budget, and pending count | agent_id |
wallet_deposit |
Add funds to a wallet | agent_id, amount, source |
wallet_withdraw |
Remove funds; rejected if monthly budget would be exceeded | agent_id, amount, destination |
wallet_transfer |
Move funds between two wallets atomically; honors source budget | from_agent_id, to_agent_id, amount, reason |
wallet_transactions |
Recent transaction history, newest first | agent_id, limit (default: 20) |
wallet_set_budget |
Set or remove monthly spending cap (0 = no limit) | agent_id, monthly_budget |
wallet_invoice |
Generate a numbered invoice record between two agents | from_agent_id, to_agent_id, amount, description |
Data Storage
~/.agentwallet/
├── wallets/
│ └── <agent_id>.json # balance, budget, total_spent
├── transactions/
│ └── <agent_id>.json # full transaction history
├── invoices/
│ └── INV-000001.json # invoice records
└── _invoice_counter.json # auto-incrementing invoice number
Budget Enforcement
When monthly_budget > 0, the server rejects any withdrawal or transfer where total_spent + amount > monthly_budget. The check runs synchronously before the balance is modified. Set monthly_budget to 0 to remove the limit.
Pricing
| Plan | Price | Included |
|---|---|---|
| Pro | $19/month | Unlimited wallets, transactions, and invoices |
License
Proprietary — see subscription terms. Source: github.com/Rumblingb/agent-wallet-mcp
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。