embedded-mcp

embedded-mcp

Lets AI assistants read, command, and debug microcontrollers over a serial connection via MCP. Provides tools for serial port discovery, reading serial output, sending commands, and decoding register values.

Category
Visit Server

README

embedded-mcp

Let an AI assistant read, command, and debug your microcontroller — over plain serial.

CI License: MIT Python 3.10+ MCP

embedded-mcp is a Model Context Protocol server. It gives an MCP-capable client (Claude Desktop, Claude Code, …) a small set of tools to talk to a real board over a serial port.

When you debug firmware, your AI pair can now see what the board prints and poke it back — instead of you copy-pasting the serial monitor by hand.

┌─────────────┐      MCP (stdio)      ┌───────────────┐     UART / USB-serial    ┌──────────────┐
│  AI client  │  ◄────────────────►   │  embedded-mcp │  ◄────────────────────►  │  your board  │
│  (Claude)   │                       │   (this repo) │       (pyserial)         │  STM32 / ...  │
└─────────────┘                       └───────────────┘                          └──────────────┘

Why

Embedded debugging is a loop of flash → watch the serial monitor → send a command → read the dump → decode a register against the datasheet. That loop is exactly the kind of tedious, context-heavy work an AI is good at — if you give it eyes and hands on the hardware. This server is those eyes and hands.

Tools

Tool What it does
list_serial_ports Discover available ports (COM5, /dev/ttyACM0, …).
read_serial Passively read what the firmware is printing for N seconds.
send_command Send a line to the device's UART shell and capture the reply.
decode_register Turn a raw value (e.g. 0x4002) into named bit-fields.
decode_register_svd Decode a register by name straight from a vendor CMSIS-SVD file.
list_flashers Report which flashers are installed (st-flash / probe-rs / openocd).
flash_firmware Flash a firmware image to the board (dry-run by default).

Install

git clone https://github.com/atillab1/embedded-mcp.git
cd embedded-mcp
pip install -e .

Use with Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "embedded-mcp": {
      "command": "embedded-mcp"
    }
  }
}

Restart Claude Desktop. Then just ask:

"List my serial ports, then read whatever COM5 is printing at 115200 for 5 seconds."

"Send status to the board on COM5 and tell me what it replies."

"Register RCC->CR read back as 0x4002. Decode it: bit 0 is HSION, bit 17 is HSEON, bit 25 is PLLRDY."

Example: decode a register

decode_register(value=0x4002, fields=[{"name":"HSION","bit":0},{"name":"HSEON","bit":17}], width=32)

{
  "value": 16386,
  "hex": "0x00004002",
  "binary": "00000000000000000100000000000010",
  "fields": [
    { "name": "HSION", "bits": "[0]",  "value": 0 },
    { "name": "HSEON", "bits": "[17]", "value": 0 }
  ]
}

Try it now (no hardware needed)

The register decoders work entirely offline — give them a quick spin:

python examples/demo.py

Flashing

flash_firmware shells out to a real flasher and is dry-run by default — it returns the exact command it would run, so you can review it before anything touches your board:

"Dry-run flashing build/app.bin to my STM32 with st-flash."

Flip dry_run=False to actually flash. Supported: st-flash (.bin), probe-rs (.elf, needs chip), openocd (.elf, needs openocd_target).

Safety notes

  • Tools open the port only for the duration of the call, then close it — they do not hold the port, so your normal IDE serial monitor can share it (one at a time).
  • Read durations are capped at 30s so a call can never hang the client.
  • flash_firmware is destructive; it defaults to a dry run and never flashes unless you explicitly pass dry_run=False.
  • This talks to whatever board is on the port. Don't point it at something you don't own.

Roadmap

  • [x] Unit tests + GitHub Actions CI
  • [x] Load a register map from an SVD file (decode_register_svd)
  • [x] Flashing hook (flash_firmware via st-flash / probe-rs / openocd)
  • [ ] Streaming/continuous monitor (notifications instead of fixed windows)
  • [ ] SVD auto-discovery from the connected chip id

Contributions welcome — open an issue or PR.

Development

pip install -e .
pip install pytest
pytest -q

License

MIT © Atilla — see LICENSE.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured