MCP server for AI agents

Connect your AI coding agent to the Xash sandbox over the Model Context Protocol. It can list products, quote and make test sales, run collections and read your recent API calls while it builds your integration.

URL: https://partners.xash.co.zw/mcp (Streamable HTTP)
Auth: Authorization: Bearer xash_test_…, a sandbox key from the portal
Sandbox only: live keys are refused, so an agent can never move real money through it.

Connect

Claude Code:

bash

claude mcp add --transport http xash-sandbox https://partners.xash.co.zw/mcp --header "Authorization: Bearer $XASH_API_KEY"

Other clients (Cursor, VS Code and others) take the same URL and header in their MCP settings:

json

{
  "mcpServers": {
    "xash-sandbox": {
      "type": "http",
      "url": "https://partners.xash.co.zw/mcp",
      "headers": { "Authorization": "Bearer xash_test_…" }
    }
  }
}

Tools

Tool	What it does
`get_account`	Your partner, the key's abilities and the sandbox balance
`list_products`	Products with prices and required recipient fields
`lookup_mobile`	The operator of a phone number and what you can sell to it
`quote`	Step 1 of a sale
`sell`	Step 2: spends a quote. Reuse `reference` and `idempotency_key` to retry safely
`get_transaction`	A sale's status, with token or PIN once completed
`collect`	Starts an EcoCash, OMari or InnBucks collection
`confirm_collection`	Submits an OMari OTP (sandbox OTP `12345`)
`get_collection`	A collection's status
`recent_requests`	Your latest sandbox API calls, to debug your integration
`send_test_webhook`	Sends a signed test event to your sandbox webhook URL
`sandbox_numbers`	The magic numbers that produce each outcome

Every tool calls the real API with your key, so results, errors and request logs are exactly what your own code will see.

MCP server for AI agents ​

Connect ​

Tools ​

MCP server for AI agents

Connect

Tools