Connect AI agents to ProxyOS via Model Context Protocol

The ProxyOS MCP server lets AI agents create cards, make purchases, and manage spending through a standardized protocol. Auto-discovery, no manual integration required.

QUICKSTART

VIEW DOCS

Server URL

https://mcp.proxyos.ai/api/mcp

Authentication

API Key

Full access — admin + agent tools

Agent Token

Agent-scoped — runtime operations only

Configuration

{
  "mcpServers": {
    "proxy": {
      "url": "https://mcp.proxyos.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer <PROXY_API_KEY>"
      }
    }
  }
}

Add this to your Claude Desktop config or MCP client settings.

Available Tools

11 tools for complete control

AGENT (6)ADMIN (5)

Agent Tools

proxy_status

Check user verification, balance, and spending eligibility

proxy_card

Create, list, freeze, close cards and get card details

proxy_intent

Declare and list spending intents

proxy_transactions

Query transaction history

proxy_onboarding

Get KYC verification and deposit links

proxy_challenges

Retrieve 3DS OTP codes

Admin Tools

proxy_users

Create and manage users

proxy_agents

proxy_policies

Create card policy templates

proxy_risk

Configure risk detection settings

proxy_webhooks

Manage webhook endpoints

MCP vs REST API

Use MCP when

[+]Your agent runs in Claude Desktop, Cursor, or MCP-compatible environments
[+]You want automatic tool discovery without manual integration
[+]You're building with frameworks that support MCP natively

Use REST API when

[+]You need fine-grained control over requests
[+]Your environment doesn't support MCP
[+]You're building a traditional backend integration

Ready to connect?

Get your first agent connected in under 5 minutes.

GET STARTED

REST API DOCS