Integration

MCP server for AI agents

Let Claude Code, Cursor or any MCP client do the integration.

/mcp speaks the Model Context Protocol (Streamable HTTP, JSON-RPC 2.0). Give your coding agent a test key and ask it to set up verification. It can create the workflow, open a test session and wire your webhook.

#Configure

Add this to .mcp.json in your project:

json

{
  "mcpServers": {
    "screenkaptain": {
      "type": "http",
      "url": "https://{{HOST}}/mcp",
      "headers": { "x-api-key": "sk_test_..." }
    }
  }
}

#Tools

Area	What the agent can do
Workflows	List, create, edit the graph, simulate, publish
Sessions	Create, list, read decisions, approve, decline, resubmit
Screening	AML screen a name, screen a wallet, search company registries
Setup	Configure webhooks, manage the blocklist

The tools call the same API with the same key, so the same permissions, limits and audit log apply.

Tip

Use a sk_test_ key for agents. Switch to live yourself once you are happy.