Wise (TransferWise) Agent Toolkit

The Wise Agent Toolkit enables the integration of various AI frameworks and libraries with Wise APIs to create and manage transfers programmatically. This library simplifies working with the Wise API and empowers developers to embed financial operations into AI-driven workflows using Python.

The toolkit supports multiple AI libraries through optional dependencies, allowing you to install only what you need.

Included below are basic instructions, but refer to the Python package documentation for more information.

Supported API Methods

This toolkit provides comprehensive access to Wise APIs through the following operations:

Quotes

• Create a quote
• Update a quote
• Get quote by ID

Recipients

• List recipient accounts
• Create recipient account
• Get recipient account by ID
• Deactivate recipient account
• Get account requirements for a quote

Transfers

• Create a transfer
• Get transfer by ID
• List transfers
• Cancel a transfer

Profiles

• List profiles
• Get profile by ID

Activities

• List activities

Supported Integrations

• ✅ LangChain - Full support with wise-agent-toolkit[langchain]
• ✅ MCP (Model Context Protocol) - Full support with wise-agent-toolkit[mcp]
• 🚧 CrewAI - Coming soon with wise-agent-toolkit[crewai]
• 🚧 AutoGen - Coming soon with wise-agent-toolkit[autogen]

Installation

Core Installation

For the basic toolkit without any AI library integrations:

pip install wise-agent-toolkit

Integration-Specific Installation

Choose your AI library and install the corresponding extra:

LangChain Integration:

pip install "wise-agent-toolkit[langchain]"

MCP (Model Context Protocol) Integration:

pip install "wise-agent-toolkit[mcp]"

Requires mcp>=2.0.0. If you're upgrading from wise-agent-toolkit <0.4.0: the mcp package made breaking changes to its server API in 2.0, so this toolkit only supports mcp 2.x going forward.

All Integrations (if you want everything):

pip install "wise-agent-toolkit[all]"

Development Installation:

pip install "wise-agent-toolkit[dev]"

Note for macOS/zsh users: The package name must be quoted to prevent shell interpretation of square brackets. Alternatively, you can escape the brackets: pip install wise-agent-toolkit\[mcp\]

Requirements

• Python 3.11+

Usage

LangChain Integration

The library needs to be configured with your Wise API key, which is available in your Wise account dashboard.

from wise_agent_toolkit.langchain.toolkit import WiseAgentToolkit

wise_agent_toolkit = WiseAgentToolkit(
    api_key="YOUR_WISE_API_KEY",
    host="https://api.transferwise.com",
    configuration={
        "actions": {
            "transfers": {
                "create": True,
            },
        }
    },
)

The toolkit works with LangChain and can be passed as a list of tools. For example:

from langchain_openai import ChatOpenAI
from langchain.agents import create_react_agent, AgentExecutor

llm = ChatOpenAI(model="gpt-4")
wise_tools = wise_agent_toolkit.get_tools()

agent = create_react_agent(
    tools=wise_tools,
    llm=llm,
    agent="zero-shot-react-description",
    verbose=True,
)

# Example usage of the agent
response = agent.run("Create a transfer of 100 EUR to John Doe's account.")
print(response)

In some cases, you will want to provide default values for specific API requests. The context parameter allows you to specify these defaults. For example:

wise_agent_toolkit = WiseAgentToolkit(
    api_key="YOUR_WISE_API_KEY",
    configuration={
        "context": {
            "profile_id": 42,
        }
    },
)

MCP (Model Context Protocol) Integration

The MCP integration allows you to expose Wise API operations as an MCP server, which can be consumed by MCP-compatible clients like Claude Desktop, Cline, or other MCP clients.

Installation for MCP Integration Only:

pip install wise-agent-toolkit[mcp]

Environment Setup: Set your Wise API credentials as environment variables:

export WISE_API_KEY="your_wise_api_key_here"
export WISE_API_HOST="https://api.transferwise.com"  # or https://api.sandbox.transferwise.tech for testing

Starting the MCP Server: Run the MCP server from the command line:

python -m wise_agent_toolkit.mcp --api_key $WISE_API_KEY --host $WISE_API_HOST

Or with explicit values:

python -m wise_agent_toolkit.mcp --api_key "your_api_key" --host "https://api.transferwise.com"

Using as an MCP Toolkit (Programmatic): If you want to integrate the MCP tools programmatically:

from wise_agent_toolkit.mcp.toolkit import WiseAgentToolkit

wise_agent_toolkit = WiseAgentToolkit(
    api_key="YOUR_WISE_API_KEY",
    host="https://api.transferwise.com",
    configuration={
        "actions": {
            "transfers": {
                "create": True,
            },
        }
    },
)

# Get MCP-compatible tools
tools = wise_agent_toolkit.get_tools()

Server Configuration: The MCP server supports the following command-line options:

• --api_key: Your Wise API key (required)
• --host: Wise API host URL (default: sandbox)
• --server_name: MCP server name (default: "wise-agent-toolkit")
• --profile_id: Wise profile ID (optional)

For production use, always use https://api.transferwise.com as the host. For testing and development, use https://api.sandbox.transferwise.tech (default).

Using with Claude Code:

Register the server for the current project:

claude mcp add wise -- python -m wise_agent_toolkit.mcp

Or check in a .mcp.json at your project root so the whole team gets it:

{
  "mcpServers": {
    "wise": {
      "command": "python",
      "args": ["-m", "wise_agent_toolkit.mcp"],
      "env": {
        "WISE_API_KEY": "${WISE_API_KEY}",
        "WISE_API_HOST": "https://api.transferwise.com"
      }
    }
  }
}

${WISE_API_KEY} is expanded from your shell environment at launch — don't hardcode the key in the file. Project-scoped .mcp.json servers need one-time approval in Claude Code, or list them under enabledMcpjsonServers in .claude/settings.local.json to skip the prompt:

{
  "enabledMcpjsonServers": ["wise"]
}

Gotcha: if you omit WISE_API_HOST, the server defaults to the sandbox API (https://api.sandbox.transferwise.tech), not your real account — set it explicitly for production data.

Verify the connection:

claude mcp list

should show wise: ... - ✔ Connected. If it instead shows Failed to connect — Connection closed, that error only means the server process exited during startup — it doesn't say why. Run the module directly to see the real traceback:

python -m wise_agent_toolkit.mcp --api_key "$WISE_API_KEY" --host "$WISE_API_HOST"

Checking Available Integrations

You can check which integrations are available in your installation:

from wise_agent_toolkit import get_available_integrations

print("Available integrations:", get_available_integrations())

Examples

For detailed examples, refer to the /examples directory in the source repository.

Adding New Integration Support

The library is designed to be extensible. To add support for a new AI library:

1. Create a new directory under wise_agent_toolkit/ for your integration
2. Implement the integration-specific toolkit and tool classes inheriting from the base classes
3. Add the optional dependency to pyproject.toml
4. Update the main __init__.py to conditionally import your integration