Build with AI (MCP)
Connect your AI coding assistant to the Verto API reference with the Verto MCP server to discover endpoints, generate integration code, and test calls in sandbox.
Build with AI (MCP)
Early accessThe Verto MCP server is in early access. We welcome your feedback — see Feedback.
The Verto MCP (Model Context Protocol) server connects AI coding assistants such as Claude, Cursor, VS Code Copilot, Codex and Gemini CLI to the published Verto API reference. Your assistant can look up endpoints, request and response schemas, and required fields directly from our documentation instead of relying on general knowledge.
The MCP server does not store or manage your credentials. If you make sandbox API calls through it, they use the credentials you configure in your own AI tool.
1. Before you start
- No credentials are needed to connect and explore the API reference.
- To make sandbox API calls you need:
- A sandbox account — see Create a sandbox account.
- Your
Client IDandAPI Key— see Access Your Credentials. - A bearer token from the login endpoint — see Authentication. Use the service-specific sandbox hosts listed in Sandbox vs production URLs.
- Build and test in sandbox first. Moving to production requires completing Go live with the Verto team.
- New to Verto? Start with Core concepts.
Keep secrets out of promptsNever paste API keys, bearer tokens, certificates or customer data into a prompt. Configure credentials in your AI tool's settings or environment variables instead.
2. Connect your AI tool
MCP server URL:
https://docs.verto.co/mcp
This is a remote MCP server using the Streamable HTTP transport. It exposes these tools:
| Tool | What it does |
|---|---|
list-specs | Lists the available Verto OpenAPI specs. |
list-endpoints | Lists API paths and HTTP methods with summaries. |
search-endpoints | Searches paths, operations and parameters to find the right endpoint. |
get-endpoint | Returns full details for an endpoint, including schemas, security and servers. |
get-server-variables | Returns the server variables for a spec. |
execute-request | Sends a request to the API (use with sandbox only). |
Choose your tool below.
Claude Code
claude mcp add --transport http verto https://docs.verto.co/mcpClaude.ai / Claude Desktop
Go to Settings → Connectors → Add custom connector and enter https://docs.verto.co/mcp as the remote MCP server URL.
Cursor
Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (project):
{
"mcpServers": {
"verto": {
"url": "https://docs.verto.co/mcp"
}
}
}VS Code (GitHub Copilot)
Add to .vscode/mcp.json in your workspace, or to your user-level MCP configuration:
{
"servers": {
"verto": {
"type": "http",
"url": "https://docs.verto.co/mcp"
}
}
}OpenAI Codex
Add to ~/.codex/config.toml:
[mcp_servers.verto]
url = "https://docs.verto.co/mcp"Gemini CLI
Add to ~/.gemini/settings.json:
{
"mcpServers": {
"verto": {
"httpUrl": "https://docs.verto.co/mcp"
}
}
}Other tools
Any MCP client that supports remote Streamable HTTP servers can connect with the same URL.
Verify your connection
Ask your assistant:
Using the Verto MCP, which endpoints do I need to create a beneficiary and send a payment?
If the answer does not reference Verto endpoints or MCP tool calls, check that the server is enabled and trusted in your tool's settings.
3. Getting better results
-
Name Verto in your prompt so the assistant uses the MCP rather than general knowledge. For example:
Using Verto, I need to convert GBP to NGN and pay a local beneficiary. What API calls do I need, and in what order? -
Add guide pages for end-to-end flows. The MCP is strongest at API reference lookups. For complete flows, paste a guide URL into your prompt alongside the MCP. For example:
Read https://docs.verto.co/docs/authentication-1 and use it with the Verto MCP to write a Node.js helper that logs in to sandbox and caches the bearer token and companyId. -
Iterate. Ask follow-up questions about required fields, error codes, and webhook payloads (Webhooks overview).
4. AI-generated content
AI output is a starting point for sandbox development, not production-ready code. Review, validate and test everything before going live.
| What to expect | What to plan for |
|---|---|
| Fast scaffolding and prototyping | Human code review |
| Endpoint discovery grounded in Verto docs | Checking results against the Verto docs and with the Verto team |
| Shorter time to your first sandbox call | Iterative prompting |
| Useful starting code | Production hardening: authentication, idempotency, webhook verification, error handling and retries |
For production concerns such as certificate-based authentication, IP whitelisting, webhook security and compliance requirements, work with the Verto team during Go live.
5. Feedback
Tell us what works and what doesn't through Contact us, or raise it with your Verto account or integration contact.
Updated 8 days ago

