Skip to main content

​
Quickstart

Get your AI agent connected to LedgerOS in minutes.

​
Prerequisites

  • A LedgerOS account with API key (lk_live_xxx)
  • An MCP-compatible client (Claude Desktop, Cursor, or mcp-remote)

​
Claude Desktop

Add to your Claude Desktop config file:
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
  "mcpServers": {
    "ledgeros": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.ledger.so/api/mcp"],
      "env": {
        "LEDGER_API_KEY": "lk_live_your_api_key"
      }
    }
  }
}
Restart Claude Desktop. You should see “ledgeros” in the MCP servers list.

​
Cursor

Add to your Cursor settings (.cursor/mcp.json): 
{
  "mcpServers": {
    "ledgeros": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.ledger.so/api/mcp"],
      "env": {
        "LEDGER_API_KEY": "lk_live_your_api_key"
      }
    }
  }
}

​
Agent Token (Runtime)

For production agents, use an agent token instead of an API key: 
{
  "mcpServers": {
    "ledgeros": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.ledger.so/api/mcp"],
      "env": {
        "LEDGER_AGENT_TOKEN": "at_live_your_agent_token",
        "LEDGER_AGENT_ID": "agent_xxx"
      }
    }
  }
}
Agent tokens are scoped to a single agent and can only access agent tools (not admin tools).

​
Test the Connection

Once connected, try these commands in your AI client: 
"Check if user user_xxx can spend"
→ Calls ledger_status

"Create a $50 card for buying office supplies"
→ Calls ledger_card with action: create

"What transactions did card_xxx have?"
→ Calls ledger_transactions

​
Verify Tools Are Available

Ask your AI: 
"What LedgerOS tools do you have access to?"
You should see a list of available tools based on your credentials.

​
Next Steps

Tools Reference

Learn each tool’s parameters

Examples

Common workflows