> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reinx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Server Overview

> MCP tools for connecting AI agents to financial infrastructure

Reinx provides 13 MCP tools for connecting AI agents to financial infrastructure. Works with Claude, ChatGPT, and any LLM that supports MCP.

## Server URL

```
https://mcp.reinx.ai/mcp
```

## Configuration

```json theme={null}
{
  "mcpServers": {
    "reinx": {
      "type": "http",
      "url": "https://mcp.reinx.ai/mcp",
      "headers": {
        "Authorization": "Bearer sk_reinx_live_your_key_here"
      }
    }
  }
}
```

Or with Claude Code:

```bash theme={null}
claude mcp add --transport http reinx https://mcp.reinx.ai/mcp \
  --header "Authorization: Bearer sk_reinx_live_your_key_here"
```

## Available tools

| Tool                      | Description                                                                                                                                                                          |
| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `check_balance`           | Live wallet holdings (per-asset + USD total). Use `get_budget_utilization` for remaining budget — under just-in-time funding the wallet balance can be low while budget is available |
| `request_wallet_payment`  | Request a wallet payment on any supported network (Base, Polygon, Arbitrum, Optimism, Solana)                                                                                        |
| `get_transactions`        | Agent's own transaction history                                                                                                                                                      |
| `get_spending_rules`      | Returns agent's spending policy                                                                                                                                                      |
| `get_denial_feedback`     | Recent denial reasons with human feedback                                                                                                                                            |
| `get_agent_status`        | Status, budget utilization, health indicator                                                                                                                                         |
| `submit_approval_request` | Request human approval for a planned purchase                                                                                                                                        |
| `get_budget_utilization`  | Budget vs actual for a time period                                                                                                                                                   |
| `list_approved_merchants` | MCC allowlist or approved merchant list                                                                                                                                              |
| `get_wallet_addresses`    | EVM and/or Solana wallet addresses                                                                                                                                                   |
| `check_merchant_allowed`  | Pre-check if a merchant is allowed                                                                                                                                                   |
| `get_spending_summary`    | Spending breakdown by category, merchant, day                                                                                                                                        |
| `check_approval_status`   | Poll for approval result on a pending request                                                                                                                                        |

<Note>
  Card-rail tools (`request_card_payment`, `request_spt`) ship with the card launch, expanding the
  surface to 16 tools.
</Note>

## Requirements

* Every tool call requires a valid API key.
* Every tool call requires a `purpose` field (3–500 characters) explaining why the agent is acting — recorded for audit and scoring.
* All calls are written to an immutable audit log.

## Go deeper

<CardGroup cols={2}>
  <Card title="Payment Workflow" icon="arrows-spin" href="/mcp-tools/payment-workflow">
    The request → poll → learn loop, idempotency, and status semantics.
  </Card>

  <Card title="Tool Reference" icon="list" href="/mcp-tools/tool-reference">
    Exact parameters and limits for every tool.
  </Card>
</CardGroup>