Connect AI Agents to UK Bank Validation
Give Claude, chatbots, and coding agents a real tool for validating UK sort codes, account numbers, and IBANs — instead of letting them guess. Built on the open Model Context Protocol and the Agent Skills standard.
Why AI Agents Need Real Bank Validation
LLMs can't validate bank details on their own
A sort code or IBAN either passes modulus checking against the official EISCD dataset or it doesn't — a language model has no way to know this from training data alone. Giving an agent a real tool call removes the guesswork and the risk of a confidently wrong answer.
Payment and onboarding copilots need it live
Support chatbots, onboarding assistants, and internal ops copilots are increasingly the first thing that touches a customer's bank details. Wiring in real-time validation stops bad data reaching a payment run before a human ever reviews it.
Coding agents can build the integration for you
Rather than reading API docs yourself, an Agent Skills-compatible coding assistant can read Mintly's skill definition directly and generate a correct client, error handling, and sandbox tests in one pass.
Your credentials stay under your control
Both integration paths keep your Mintly API key in a local environment variable. The agent only ever sees a validation result — never your key, and Mintly does not persist the account numbers you check.
Two Ways to Connect
Pick the integration that matches what you're building — a live agent that answers questions, or a coding assistant that writes the integration for you.
MCP Server
Run a lightweight local MCP server that exposes Mintly's bank validation as a tool. Connect it to Claude Desktop, Chat GPT, or any MCP-compatible AI agent, and ask questions in plain English.
- For runtime chatbots and AI assistants
- Works with Claude Desktop and any MCP client
- 12-minute setup guide, Python + uv
Agent Skill
Download the mintly-api Agent Skill and drop it into your coding assistant's skills folder. Invoke /mintly-api to generate client code, handle errors, and set up sandbox tests.
- For AI coding assistants, not runtime agents
- Claude Code, Copilot, Cursor, Codex, Gemini CLI
- Built on the open Agent Skills standard
See It In Action
A few lines of config connect an agent to live UK bank validation.
Full walkthrough, including the Python server code, is in the MCP server guide.
Where Teams Use This
Payment onboarding chatbots
An AI assistant collecting bank details during onboarding can validate them in the same conversation, catching typos before the customer moves on.
Support agents verifying bank details
Customer support copilots can confirm a sort code and account number are valid — and which payment schemes they support — without leaving the chat.
Internal ops copilots
Ask an internal agent to validate a batch of supplier or payroll accounts before a payment run, flagging anything that fails modulus checking.
Coding agents building your integration
Point Claude Code, Copilot, or Cursor at the mintly-api skill and have it write the client, tests, and error handling for your project.
Frequently Asked Questions
What is the Model Context Protocol (MCP)?
MCP is an open standard for connecting AI systems to external tools and data sources. It lets an AI agent — such as Claude Desktop or a custom LLM application — call a defined set of functions ("tools") instead of guessing an answer. Mintly's MCP guide shows you how to run a local MCP server that exposes UK bank validation as a tool any MCP-compatible agent can call.
What's the difference between the MCP server and the Agent Skill?
The MCP server is for runtime AI agents and chatbots — it gives an assistant like Claude Desktop a live tool to validate a sort code or account number during a conversation. The Agent Skill is for AI coding assistants — it gives tools like Claude Code, GitHub Copilot, or Cursor in-context knowledge of the Mintly API so they can write correct integration code for you while you build.
Which AI tools can connect to Mintly?
The MCP server works with Claude Desktop and any MCP-compatible client. The Agent Skill works with Claude Code, VS Code with GitHub Copilot, Cursor, OpenAI Codex, and Gemini CLI — any client built on the open Agent Skills standard.
Is it safe to let an AI agent check bank account details?
Yes, when set up correctly. Your Mintly API key stays in a local environment variable, never in a prompt or in source control. The MCP server acts as a proxy — the agent only ever sees the validation result, not your credentials — and Mintly does not persist the account numbers submitted for checking.
Do I need a paid Mintly plan to use the MCP server or Agent Skill?
No. You can start with the free trial API key, so you can build and test against the sandbox and real API before upgrading. Paid plans raise your rate limits and request volume for production use. See pricing for details.
Give Your Agent a Real Tool
Free sandbox access, no credit card required. Set up the MCP server or download the Agent Skill in minutes.
Prefer to talk it through first? Contact us.