Skip to content

OnStepNinja/AiBridgeMCP

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

46 Commits
 
 
 
 
 
 

Repository files navigation

🚀 LUNA v5.9.1b Released — Lua scripting engine on ESP32. Autonomous observation control, ASCOM/Alpaca integration, and more. ➡️ github.com/OnStepNinja/LUNA


AiBridgeMCP: The World's Best AI Bridge for Legacy Hardware

"Stop writing code. Just give the manual to the AI."

AiBridgeMCP is a standalone physical bridge that connects any legacy device with an RS-232C port directly to modern AI agents (like Claude). From 80s microcomputers to expensive industrial instruments and telescope mounts — tasks that once required expert programming are now transformed into simple "conversations with AI."

LUNA — The Next Evolution

LUNA extends AiBridgeMCP with a Lua scripting engine running directly on the ESP32. Claude can deploy scripts onto the chip, which then run autonomously — no PC, no cloud, no further AI involvement required. Lua scripts can also control any ASCOM/Alpaca-compatible device — focusers, filter wheels, rotators, weather stations — directly over HTTP, without a Windows PC or ASCOM Platform.

Designed for astronomical instrument control: Takahashi Temma2, NS-5000, and OnStep NS-3000.


💰 The $5 Revolution: A New World with a Single ESP32

The most shocking part of this project is its overwhelming cost-performance.

  • A Tiny $5 Investment: All you need to realize this revolutionary AI system is a single, off-the-shelf ESP32 device (approx. $5).
  • Standalone & PC-Free: No expensive servers or dedicated PCs are required. This tiny chip functions as an independent, intelligent AI bridge.
  • Redefining Your Assets: Your "mysterious old equipment" gathering dust can be reborn as a cutting-edge AI-driven system for the price of a cup of coffee.

💡 Three Innovations of AiBridgeMCP

1. Chat Across Decades with 80s Micros

Turn your Apple II, MSX, or CP/M machine into a terminal for modern AI. Chat in real-time with the intelligence of the 21st century directly from 40-year-old hardware via RS-232C.

2. Instant AI Systems: Manuals Instead of Programs

Traditionally, automating old equipment took weeks of reading manuals and writing complex code. With AiBridgeMCP, you simply provide the manual (PDF, etc.) to the AI. The AI understands the commands and operates the device on your behalf.

3. Natural Language Control for Professional Gear

Supports astronomical GOTO systems (OnStep NS-3000, NS-5000), Digital Multimeters (DMM), and Oscilloscopes. Just say "Measure the voltage" or "Go to that star" in plain Japanese or English to execute complex operations.


📈 A Massive Paradigm Shift

Feature Traditional Method AiBridgeMCP + AI
Preparation Weeks to Months Minutes (Just provide the manual)
Required Skill Expert Programming Natural Language Conversation
Cost Thousands in Dev Fees Just $5 for Hardware

📚 Documentation

Setup guides, technical references, and FAQ. Ask questions directly inside NotebookLM and get answers from the official documentation. 👉 https://notebooklm.google.com/notebook/f2ce997f-18c2-44c4-8738-973b204c190c

💬 Community — Join our Facebook group to ask questions, share your results, and connect with other users. The developer is also a member. 👉 https://www.facebook.com/groups/1230935959149731


🛠 Technical Specifications (Overview)

  • Hardware: ESP32 (Standalone operation)
  • Interface: RS-232C (TTL level, requires converter)
  • Protocol: Model Context Protocol (MCP) — Streamable HTTP
  • Core Tools: serial_query, serial_write, get_system_info

📜 License & Copyright

  • License: MIT License
  • Copyright: (C) 2026 Nishioka Sadahiko
  • Disclaimer: This system is provided "as is". The author is not responsible for any damage to equipment resulting from its use.

Developer: Nishioka Sadahiko (OnStepNinja)

Blog: https://nskikaku.blog.fc2.com/


📋 Release History

LUNA v5.9.1b — 2026-03-25

LUNA (v5.9.1b) adds a Lua scripting engine to the AiBridgeMCP server. Claude can now deploy scripts directly onto the ESP32, which then run autonomously — no PC, no cloud, no further AI involvement required.

Designed for astronomical instrument control: Takahashi Temma2, NS-5000, OnStep NS-3000.

➡️ github.com/OnStepNinja/LUNA


Streamable HTTP — 2026-02-27

AiBridgeMCP previously supported SSE (Server-Sent Events), and now it has successfully connected using Streamable HTTP as well!! Version 2.0.0 🚀


⚠️ Known Issue: MCP Instability with Claude Desktop v1.1.3189

Important Notice for AiBridgeMCP Users

If you are experiencing unstable MCP connections or tool call failures, Claude Desktop v1.1.3189 (released February 17, 2026) contains known bugs that may be the root cause — not AiBridgeMCP itself.

What's happening

  • MCP regression: The "searched available tools" behavior is broken in this version, causing unreliable tool discovery and invocation.
  • UI lag / system overload: A Hyper-V virtual machine introduced for the Cowork feature starts automatically at every launch, even when unused. This consumes significant CPU, memory, and I/O in the background, degrading all MCP communication.

Solution

Update Claude Desktop to v1.1.4088 (released February 24, 2026) or later.

AiBridgeMCP has been verified to operate correctly on stable versions of Claude Desktop. If you continue to experience issues after updating, please open an issue in this repository.

References


📦 v1.8.8 All-in-One Package — 2026-02-24

Includes everything you need to get started immediately.

🛒 Official BOOTH Store

Get licenses, detailed design data, and upcoming official hardware.


Notice: Transition to Streamable HTTP — 2026-03-13

To align with Model Context Protocol (MCP) version 2026-03-26, the ecosystem transitioned from legacy HTTP+SSE transport to Streamable HTTP.

  • Key Change: Legacy SSE transport deprecated. AiBridgeMCP supports the new Streamable HTTP specification.
  • Benefits: Improved session recovery and better compatibility with various network environments.
  • Action Required: Please use the latest version of AiBridgeMCP.

MCP Inspector 画面