Hi HN,

I built PolyMCP, an open-source framework around the Model Context Protocol (MCP) that lets you expose existing Python functions as AI-callable tools — without rewriting them or adopting a custom SDK.

The goal is simple: If you already have working Python code, you should be able to make it accessible to LLM agents in minutes.

What it does

PolyMCP introspects regular Python functions and exposes them as MCP tools automatically. No decorators required. No framework lock-in.

It grew into a small ecosystem: • PolyMCP (core) – Turn Python functions into MCP tools • PolyMCP Inspector – A visual UI to browse, test, and debug MCP servers • MCP SDK Apps – A lightweight way to build AI-powered apps with tools + UI resources

Why I built this

While experimenting with MCP and AI agents, I found that integrating existing codebases was often the painful part. Most solutions require rewriting logic around a specific SDK or heavily annotating functions.

PolyMCP focuses on: • Minimal intrusion into existing code • Clean separation between business logic and AI tooling • Easy debugging via a visual inspector

Example use cases • Expose internal APIs or legacy scripts to LLM agents • Automate operational workflows • Build internal copilots over real systems • Prototype AI agents that interact with production services

Works with OpenAI, Anthropic, and Ollama (including local models).

It’s still evolving and I’m actively iterating. I’d really appreciate feedback — especially from people building agents or experimenting with MCP in production environments.

GitHub: • Core: https://github.com/poly-mcp/PolyMCP • Inspector: https://github.com/poly-mcp/PolyMCP-Inspector • SDK Apps: https://github.com/poly-mcp/PolyMCP-MCP-SDK-Apps

Happy to answer technical questions.

Deepwiki.com (Devin) documentation of Sutskever-30-implementations

Tékumel

Reports of Telnet's Death Have Been Greatly Exaggerated

Agentic Engineering

WebMCP started as a solution to auth for agents at Amazon

Ford Falls Behind China's BYD in Global Sales for the First Time

Show HN: agent alcove – Claude, GPT, and Gemini debate across forums

An ice dance duo skated to AI music at the Olympics

Musk says xAI was reorganized, leading to some layoffs

Show HN: CodeMoot – Bridge Between Claude Code and Codex CLI

Today is my last day at Anthropic. I resigned

Show HN: PolyMCP – Expose Python functions as MCP tools

Ghost CMS adds support for welcome emails

DOJ says Trenchant boss sold exploits to Russian broker

PXL2000

Video: A Brief Overview of the Atari 800XL

Show HN: Yet another music player but written in Rust

Apple's Latest Attempt to Launch the New Siri Runs into Snags

GLM 5 – The Next-Gen Frontier

Iran's Digital Surveillance Machine Is Almost Complete

Deleted doesn't mean gone: How police recovered Nancy Guthrie's doorbell footage

A few design decisions for a new chat platform

Apple withholds 18.7.5 security update from iPhones and iPads supporting iOS 26

Building Modern Databases with the FDAP Stack

TypeScript 6.0 Beta

Three Conversations Worth Having with Your CTO

Microwave Oven Failure: Spontaneously turned on by its LED display (2024)

Show HN: Codingagents.md – The open directory for AI coding agents

You Don't Need It

Singing the Gospel of Collective Efficacy

Show HN: Open-source monitoring for AI agents (MCP-compatible)