MCP Server
Give AI agents structured access to Easy Labs through the Model Context Protocol.
The Easy Labs MCP Server lets AI agents — Claude Desktop, custom assistants, hosted agent platforms — read your Easy Labs data through structured tools and resources instead of raw HTTP calls. It speaks the Model Context Protocol, an open standard for connecting LLMs to external systems.
npx -y @easylabs/mcp-serverWhat it gives you
Tools
Read-only tools for transfers, disputes, settlements, customers, and balance transfers.
Resources
Merchant profile and summary surfaces for grounding context.
Transports
Stdio (default, for local clients) and HTTP/SSE-compatible (for remote/daemonized usage).
When to use the MCP server
- You want Claude Desktop (or any MCP-compatible client) to inspect your Easy Labs activity.
- You're building an AI agent that needs structured, validated payment data — not raw API responses.
- You need a drop-in surface for tool-calling LLMs without writing custom integrations per provider.
For server-to-server integrations, prefer @easylabs/node directly — it's the canonical SDK and has the full surface. The MCP server today exposes a subset of that surface, packaged for AI consumption.
Quickstart
Roadmap
The MCP server today is read-only and covers transfers / disputes / settlements / customers / balance-transfers. Modernization work tracked at easy-mcp modernization plan will:
- Expand coverage to the full
@easylabs/nodesurface (subscriptions, invoices, payouts, recipients, embedded checkout, etc.) - Add write tools with confirmation flows (
create_transfer,create_subscription,send_invoice, etc.) - Ship sandbox/production env splits matching the SDK
- Auto-generate tools from the OpenAPI spec so coverage stays in lockstep