Interswitch MCP Server

Interswitch MCP Server

A Model Context Protocol (MCP) server that enables AI assistants to interact with Interswitch APIs for payments, transfers, VAS, cardless paycodes, Transaction Search, Card 360, lending, payouts, agency banking, and fintech card-processing utilities.

[!WARNING] Public Preview: This MCP server is currently in public preview. It supports both Interswitch sandbox and production environments, so use cautiously, start with read-only mode, and report any issues you encounter.

Quick Start

Install and run via npm (recommended):

npx interswitch-mcp-server

Or install globally:

npm install -g interswitch-mcp-server
interswitch-mcp-server

Or for local development, clone and build:

git clone https://github.com/Dairus01/Interswitch-mcp-server.git
cd Interswitch-mcp-server
npm install
npm run build

Then configure your MCP client to use the server (see Client Integration).

Requirements

• Node.js v18+
• npm or yarn
• Interswitch API credentials for the domains you want to use
• Required Interswitch merchant, payable, initiating entity, terminal, or domain-specific identifiers for your use case

Configuration Options

You can select the API environment with INTERSWITCH_ENV:

1. Sandbox (default): INTERSWITCH_ENV=sandbox
2. Production: INTERSWITCH_ENV=production

Security note: Production mode uses real Interswitch production endpoints. If INTERSWITCH_ENV=production, valid production credentials are configured, INTERSWITCH_READ_ONLY=false, and a write tool is called with confirm: true, the server can make real production API calls.

For safer first-time setup, start in read-only mode:

INTERSWITCH_ENV=production \
INTERSWITCH_READ_ONLY=true \
INTERSWITCH_CLIENT_ID=your_production_client_id \
INTERSWITCH_CLIENT_SECRET=your_production_client_secret \
interswitch-mcp-server

When you are authorized and ready to allow live writes, set:

INTERSWITCH_READ_ONLY=false

Client Integration

The Interswitch MCP Server works with any MCP-compatible client. Below is the standard configuration schema used by most clients (Claude Desktop, ChatGPT Desktop, Cursor, Windsurf, VS Code, Claude Code, etc.).

Using npm (recommended)

For npm-installed server:

{
  "mcpServers": {
    "interswitch": {
      "command": "npx",
      "args": ["-y", "interswitch-mcp-server"],
      "env": {
        "INTERSWITCH_ENV": "sandbox",
        "INTERSWITCH_READ_ONLY": "true",
        "INTERSWITCH_CLIENT_ID": "your_client_id",
        "INTERSWITCH_CLIENT_SECRET": "your_client_secret",
        "INTERSWITCH_MERCHANT_CODE": "your_merchant_code",
        "INTERSWITCH_PAYABLE_CODE": "your_payable_code",
        "INTERSWITCH_INITIATING_ENTITY_CODE": "your_entity_code"
      }
    }
  }
}

For production, change the environment values intentionally:

{
  "mcpServers": {
    "interswitch": {
      "command": "npx",
      "args": ["-y", "interswitch-mcp-server"],
      "env": {
        "INTERSWITCH_ENV": "production",
        "INTERSWITCH_READ_ONLY": "true",
        "INTERSWITCH_CLIENT_ID": "your_production_client_id",
        "INTERSWITCH_CLIENT_SECRET": "your_production_client_secret",
        "INTERSWITCH_MERCHANT_CODE": "your_production_merchant_code",
        "INTERSWITCH_PAYABLE_CODE": "your_production_payable_code",
        "INTERSWITCH_INITIATING_ENTITY_CODE": "your_production_entity_code"
      }
    }
  }
}

Using a local build

If you've cloned and built the server locally:

{
  "mcpServers": {
    "interswitch": {
      "command": "node",
      "args": ["/path/to/Interswitch-mcp-server/dist/index.js"],
      "env": {
        "INTERSWITCH_ENV": "sandbox",
        "INTERSWITCH_READ_ONLY": "true",
        "INTERSWITCH_CLIENT_ID": "your_client_id",
        "INTERSWITCH_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

[!IMPORTANT] When setting command: "node", you should ensure you're using Node v18+. If you are using a package manager, you might need to get the path of your Node binary by running this command in your CLI:

Linux and MacOS

which node

Windows

where node

Once you have the path, use it as the value of the MCP Server command in the JSON configuration. e.g., command: "path/to/installation/bin/node"

Where to add this configuration

How It Works

The Interswitch MCP Server exposes Interswitch API domains to AI assistants through dedicated MCP tools. The server does not parse an OpenAPI file at runtime; instead, it provides curated, typed tools for the supported Interswitch domains.

Architecture Diagram

graph TD
    subgraph MCP Server
        A[Interswitch MCP Server]
        B[Tool Registry]
        C[Resource Registry]
    end

    subgraph Auth Layer
        D[OAuth2 Passport Manager]
        E[Card 360 Token Manager]
        F[Transaction Search Manager]
        G[Agency Banking Token Manager]
        H[Legacy Auth Manager]
    end

    subgraph Core Services
        I[Payments Service]
        J[VAS Services]
        K[Transfers Service]
        L[Cardless Services]
        M[Transaction Search Service]
        N[Card 360 Service]
        O[Lending Service]
        P[Transfer Service]
        Q[Payouts Service]
        R[Agency Service]
        S[Fintech Utilities]
    end

    subgraph Safety Layer
        J1[Confirmation Gate]
        J2[Read-Only Mode]
        J3[Redaction System]
        J4[Audit Logging]
    end

    A --> B
    A --> C
    A --> D
    A --> E
    A --> F
    A --> G
    A --> H
    B --> I
    B --> J
    B --> K
    B --> L
    B --> M
    B --> N
    B --> O
    B --> P
    B --> Q
    B --> R
    B --> S
    I --> J1
    J --> J1
    K --> J1
    L --> J1
    M --> J1
    N --> J1
    O --> J1
    P --> J1
    Q --> J1
    R --> J1
    S --> J1
    J1 --> J2
    J2 --> J3
    J3 --> J4
    J4 --> Z[External APIs]
    
    style MCP Server fill:#f0f8ff,stroke:#333,stroke-width:2px
    style Auth Layer fill:#e6f7e6,stroke:#333,stroke-width:1px
    style Core Services fill:#f9f9f9,stroke:#666,stroke-width:1px
    style Safety Layer fill:#fff0f0,stroke:#c33,stroke-width:1px
    style Z fill:#d4d4d4,stroke:#999,stroke-dasharray: 5 5

Each tool:

1. Validates its input with a strict schema
2. Applies confirmation and read-only safety gates
3. Selects the configured sandbox or production base URL
4. Uses the correct authentication system for the domain
5. Calls the Interswitch API only when the tool is invoked
6. Redacts sensitive fields from returned data
7. Returns a normalized MCP response

Authentication systems

Available Tools

Available Resources

Example

When you ask your AI assistant something like "Use Interswitch to initialize a 5000 NGN payment for customer@example.com", here's what happens behind the scenes:

1. The assistant selects isw_initialize_payment
2. It passes explicit money input such as amountNaira: 5000
3. For write operations, it must include confirm: true
4. The server validates the request, converts the amount safely where required, authenticates with Interswitch, and calls the configured sandbox or production endpoint
5. You get a normalized response with redacted raw data

Prompt recommendation

To get the best results when using this MCP server, be specific in your prompts and always include "Interswitch" in your requests.

Good prompts:

• "Initialize an Interswitch payment for 50000 NGN for user@example.com."
• "Check the Interswitch transaction status for reference order-1001."
• "Resolve Interswitch bank account 0123456789 with bank code 058."
• "Send an Interswitch single transfer for 1000 NGN with confirm true."
• "Read the Interswitch response code resource."

Less effective prompts:

• "List my transactions" (unclear which service and reference/search criteria to use)
• "Send money" (missing beneficiary, bank, amount, and confirmation details)
• "Pay this bill" (missing biller, package, customer, and amount details)

Being explicit ensures the LLM narrows down to the right tool quickly and reduces ambiguity.

Production Safety

This server can route to production Interswitch endpoints when INTERSWITCH_ENV=production is set. Production readiness depends on correct credentials, Interswitch account permissions, and your organization's operational controls.

Safety behavior built into the server:

• Write tools require confirm: true
• High-risk sensitive read tools require confirm: true
• INTERSWITCH_READ_ONLY=true blocks all writes before any external API call
• Low-risk reads do not require confirmation
• Raw responses are redacted before being returned
• Tokens and secrets are never intentionally returned in tool output
• Card PIN payloads are encrypted locally when the required RSA public key is configured
• Production startup warnings are written to stderr instead of silently hiding risky configuration

Recommended production rollout:

1. Configure production credentials but keep INTERSWITCH_READ_ONLY=true
2. Verify startup warnings and config summary
3. Test safe read-only tools first
4. Confirm domain credentials for the exact tools you intend to use
5. Switch INTERSWITCH_READ_ONLY=false only when live writes are authorized
6. Require human review before prompts that include confirm: true

Development

Run locally (without building)

For local development and testing, you can run the TypeScript source directly:

npm run dev

Run with MCP Inspector

npm run inspect

Build

npm run build

Run tests

npm test

Type-check

npm run lint

Preview npm package contents

npm pack --dry-run

Troubleshooting

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

See CONTRIBUTING.md for more details.

License

MIT