MCP and AI coding tools

Everything on this site is available to coding assistants in three forms. Pick the one your tool supports.

Form	Where	What it gives an assistant
Plain text	docs.mesta.xyz/llms.txt, and any page with `.md` added to its address	Every guide and reference page as markdown
Method reference	reference.md for TypeScript, reference.md for Python	Every SDK method with a runnable example
MCP	`https://docs.mesta.xyz/mcp`	Search the guides, read any endpoint's definition, and run requests against the staging API with your sandbox key

Connect over MCP

The MCP endpoint serves this documentation and the API reference over the Model Context Protocol. No account is needed to read. Every client below takes the same address, https://docs.mesta.xyz/mcp; the headers are optional and only let the assistant execute requests, so use sandbox keys from the Sandbox Portal. The endpoint reaches the staging API only, never production.

claude mcp add --transport http mesta-docs https://docs.mesta.xyz/mcp

With sandbox keys, so the assistant can also run requests:

claude mcp add --transport http mesta-docs https://docs.mesta.xyz/mcp \
  -H "x-api-key: YOUR_SANDBOX_API_KEY" -H "x-api-secret: YOUR_SANDBOX_API_SECRET"

The Codex CLI, the IDE extension and the ChatGPT desktop app share one configuration.

codex mcp add mesta-docs --url https://docs.mesta.xyz/mcp

With sandbox keys, in ~/.codex/config.toml:

[mcp_servers.mesta-docs]
url = "https://docs.mesta.xyz/mcp"
http_headers = { "x-api-key" = "YOUR_SANDBOX_API_KEY", "x-api-secret" = "YOUR_SANDBOX_API_SECRET" }

In .cursor/mcp.json for one project, or ~/.cursor/mcp.json for all of them:

{
  "mcpServers": {
    "mesta-docs": {
      "url": "https://docs.mesta.xyz/mcp",
      "headers": {
        "x-api-key": "YOUR_SANDBOX_API_KEY",
        "x-api-secret": "YOUR_SANDBOX_API_SECRET"
      }
    }
  }
}

In ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "mesta-docs": {
      "serverUrl": "https://docs.mesta.xyz/mcp",
      "headers": {
        "x-api-key": "YOUR_SANDBOX_API_KEY",
        "x-api-secret": "YOUR_SANDBOX_API_SECRET"
      }
    }
  }
}

gemini mcp add --transport http mesta-docs https://docs.mesta.xyz/mcp

With sandbox keys:

gemini mcp add --transport http mesta-docs https://docs.mesta.xyz/mcp \
  --header "x-api-key: YOUR_SANDBOX_API_KEY" --header "x-api-secret: YOUR_SANDBOX_API_SECRET"

Any other client that speaks the protocol connects the same way: a streamable HTTP server at that address, with optional headers.

What the assistant can do

search-endpoints, list-endpoints and get-endpoint: find an operation and read its parameters, responses and security scheme, straight from the API reference.
search and fetch: search and read the guides, including the workflow guides that give the order of calls for onboarding a sender or making a payout.
execute-request: run a request against the staging API with the keys in your configuration. A quote is a safe first call; it prices a transfer and moves no money.

📘
For building, not for operating
This endpoint is for writing an integration. An agent that operates on Mesta at runtime, requesting quotes and payments under limits a person has approved, uses the Mesta agent platform described at mesta.xyz/agentic-payments.

Once the integration exists, the Official SDKs are the typed way to call the API from your code.