A comprehensive Model Context Protocol skill that enables AI agents to autonomously request virtual payment cards and manage transactions within secure spending limits.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install proxy-payments
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install proxy-payments using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Proxy Pay MCP?
Proxy Pay MCP is a specialized tool designed to provide a secure financial layer for AI agents. By integrating this skill, developers can grant their agents the ability to perform real-world financial transactions, such as paying for cloud services, purchasing digital assets, or managing subscriptions. The skill acts as a bridge between high-level agent logic and financial infrastructure, ensuring that all spending is governed by strict, user-defined policies. Within the ecosystem of Openclaw Skills, Proxy Pay MCP stands out as a critical component for building agents that need to operate independently in commercial environments.
This skill provides agents with the capability to declare payment intents, which are then evaluated against spending policies. If the request aligns with the predefined limits, a virtual card is issued instantly. This architecture ensures that even fully autonomous agents remain under financial control, with multi-layered protections including merchant restrictions and human-in-the-loop approval thresholds for high-value purchases. For developers using Openclaw Skills, this provides a production-ready solution for agentic commerce.
Proxy Pay MCP Use Cases
- Managing and paying for recurring SaaS subscriptions and API usage fees.
- Provisioning on-demand cloud resources like AWS or Google Cloud for specific tasks.
- Handling procurement tasks where agents need to buy digital assets or research materials.
- Automating expense reconciliation by matching transactions to specific agent intents.
- Implementing budget-capped research projects where AI agents manage their own funding allocations.
How Proxy Pay MCP Works
- The AI agent generates a payment intent using the
proxy.intents.createtool, specifying the amount and merchant. - The system evaluates the intent against the spend policy assigned to the agent, checking for limits and restrictions.
- If the intent is within policy, a virtual card is auto-provisioned; otherwise, it moves to a pending state for human approval.
- Once the intent is approved, the agent retrieves the sensitive card details (PAN, CVV) to complete the checkout process.
- The transaction is tracked in real-time, and the agent uses the
proxy.receipts.attachtool to upload evidence for auditing. - All financial activity is reconciled against the initial intent to provide a transparent audit trail within Openclaw Skills.
Proxy Pay MCP Setup
- Sign up for a Proxy account at useproxy.ai and complete the required KYC/KYB verification.
- Create a Spend Policy and an Agent in the Proxy dashboard to define spending boundaries.
- Authenticate using Claude Code by running:
claude mcp add proxy --transport http https://mcp.useproxy.ai/api/mcp
- Extract your access token from
~/.claude/.credentials.jsonand add it to your environment asPROXY_TOKEN. - Update your
config/mcporter.jsonfile to include the Proxy MCP server configuration:
{
"mcpServers": {
"proxy": {
"baseUrl": "https://mcp.useproxy.ai/api/mcp",
"headers": {
"Authorization": "Bearer ${PROXY_TOKEN}"
}
}
}
}
- Confirm the setup by calling
mcporter list proxyto ensure all tools are available.
Proxy Pay MCP Data Schema & Taxonomy
Proxy Pay MCP organizes data into several entities to ensure clear financial reporting and compliance for Openclaw Skills users.
| Entity | Key Attributes | Purpose |
|---|---|---|
| Payment Intent | intentId, amount, status, merchant |
Records the agent's request to spend money. |
| Virtual Card | cardId, last4, expiry, status |
The issued payment instrument used for checkout. |
| Policy | limits, merchant_allowlist, thresholds |
The governance rules applied to all transactions. |
| Transaction | transactionId, amount, timestamp |
The actual ledger entry from the banking network. |
| Receipt | url, intentId, description |
The digital evidence linked to a specific transaction. |
Proxy Pay MCP Advanced Features
- Policy Simulation: Allows agents to dry-run an intent to see if it would be approved before actually creating it.
- Agent-Scoped Tokens: Use specific agent tokens for autonomous environments to ensure the agent only accesses its own data.
- Merchant Intelligence: Tools like
proxy.merchants.resolvehelp agents identify and allowlist canonical merchant entities. - Human-in-the-Loop Hooks: Programmatic endpoints for manual approval or rejection of high-stakes transactions.
- Multi-Currency Support: Handles transactions and funding in various currencies, including ACH, wire, and USDC on Solana.
SKILL.md
Loading
Related Openclaw Skills
Proxy Pay
A fintech automation skill for Openclaw Skills that enables instant creation of payment intents and provisioning of secure virtual cards via simple commands.
Proxy Pay MCP
A secure Model Context Protocol server that enables AI agents to issue virtual cards, manage payment intents, and track expenses autonomously.
Proxy MCP Integration
A Model Context Protocol server that empowers AI agents to manage payments, issue virtual cards, and track financial transactions autonomously.
Proxy Balance Tracker
Monitor Proxy account balances and manage pending transaction intents in real-time.
Proxy Status
A robust diagnostic tool for tracking the lifecycle of payment intents, card issuances, and transaction matching within the Proxy ecosystem.
QELT Blockchain
A high-performance interface for querying and interacting with the QELT Layer-1 blockchain via standard JSON-RPC.
Featured*
