SECCI Pro Developer API & MCP Server

SECCI.pro offers a REST API and a public Model Context Protocol (MCP) server for programmatic SECCI generation and APRC calculation.

MCP server

• URL: https://obzapshhsmzencudrbis.functions.supabase.co/mcp
• Transport: streamable HTTP
• Authentication: none (public)
• Manifest: /.well-known/mcp.json
• Tools: compute_aprc, compute_overdraft_aprc, build_secci_url

An LLM-friendly summary of the whole site (regulatory basis, APRC formula, MCP tools) is available at /llms.txt.

Full setup instructions for Claude, ChatGPT, Cursor and VS Code are on the MCP server page.

Installation

There is no SDK to install: the REST API is plain HTTPS and JSON. Integration is three steps — create an API key, store it as the SECCI_API_KEY environment variable, and send it as Authorization: Bearer <key>. The developer page carries complete, copy-paste client files for Node/TypeScript, Python, PHP and curl, plus a verification request whose expected APRC is 10.7372 %.

• Node 18+ — built-in fetch, no dependencies.
• Python 3.9+ — pip install requests.
• PHP 8.0+ — ext-curl.
• No-code — paste the https://secci.pro/widget.js script tag.
• AI agents — claude mcp add --transport http secci-pro (no key required).

Install it with an AI coding assistant

A machine-readable install guide for AI coding assistants (Lovable, Cursor, Claude Code, GitHub Copilot agent mode, Windsurf) is published at /llms-install.txt. It tells the assistant which of the three integrations to choose, gives a reference client, and includes an acceptance test with a known expected APRC, plus the rules it must follow: never expose the API key in browser code, never reimplement the APRC formula, always keep a human review step before a SECCI is issued. The developer page carries ready-made prompts for adding the calculator to a website, wiring the REST API into an app backend, connecting your own AI assistant, and auditing existing hand-written APRC code.