apcore-mcp

Zero-code bridge for MCP, OpenAI Tools, and JWT Auth

About

apcore-mcp is a high-performance adapter that bridges apcore projects to AI protocols with zero intrusion. It automatically turns modules into secure tools via MCP (Model Context Protocol) or OpenAI-compatible JSON definitions. With built-in JWT Authentication and Elicitation-based Approval support, it provides a complete security layer for AI tool execution. The project includes an interactive Tool Explorer UI for live testing, a Streaming Bridge for real-time outputs, and intelligent schema transformation including $ref resolution. It ensures zero-boilerplate integration for Django, NestJS, Flask, and TipTap applications.

Features

Zero Intrusion: Bridge existing projects to AI agents without code changes

Dual Support: Simultaneous exposure as MCP Server and OpenAI Tools

JWT Authentication: Built-in secure token validation and identity injection

Tool Explorer UI: Interactive browser-based testing and documentation suite

Streaming Bridge: Real-time notification and chunk accumulation support

User Elicitation: Interactive input requests during module execution

Intelligent Conversion: Automated schema transformation and $ref inlining

Annotation Mapping: Direct mapping of behavioral hints to MCP annotations

Sanitized Execution: Secure error handling with AI suggestions and no leaks

Dynamic Registration: Real-time module registration reflected in AI tools

Get Started

Python Implementation

Python reference implementation of apcore-mcp. 260+ tests, 90%+ coverage.

View Repository PyPI Package

Install

pip install apcore-mcp

Quick Start

from apcore import Registry
from apcore_mcp import serve

# Auto-discover all modules
registry = Registry(extensions_dir="./extensions")
registry.discover()

# Start MCP server (stdio by default)
serve(registry)

# Or export as OpenAI tools
from apcore_mcp import to_openai_tools
tools = to_openai_tools(registry)