GR-MCP: A gnuradio MCP Server
LocalAbout GR-MCP: A gnuradio MCP Server
Marconi is an LLM-driven RF tool for Claude Code that lets users describe radio tasks in plain language. It surveys spectrum, builds and runs receivers, and saves reproducible outputs such as SigMF captures, YAML pipelines and .grc flowgraphs. Version 1.0 is simulation-only and does not yet support hardware.
Key features
- 01Spectrum survey with PSD, detection and spectrogram
- 02Simulated device registration from a description
- 03Receiver build, validation, run and verification
- 04SigMF captures and YAML pipeline output
- 05Reproducible .grc flowgraph project files
Use cases
- 01Analyzing an existing IQ recording to decode its strongest carrier
- 02Building a simulated FM broadcast receiver
- 03Diagnosing a receiver that produces no output
MCP support
The project runs as an MCP server in Claude Code, started on first use by uv, and loads its skills alongside it.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMarconi 🤖
o)))
|
.-------.
| o o |
| v |
'-------'
Hi! I'm Marco.
LLM-driven RF for Claude Code. Describe what you want on the air and I survey the
spectrum, build and run the receiver, look at the signal, and leave a reproducible
RF project behind — SigMF captures, YAML pipelines, and .grc flowgraphs you own.
Early development — v1.0 is simulation-only (no hardware yet). See
ROADMAP.md.
Proviously known as GNURadio/GR-MCP.
Requirements
uvand Python ≥ 3.13- GNU Radio 3.10+ installed system-wide (conda-forge, your distro's package manager, or Homebrew) — needed to run pipelines and simulate. Analyzing existing IQ files doesn't require it.
Install
In Claude Code:
/plugin marketplace add yoelbassin/gr-mcp
/plugin install marconi
That's it — uv starts the MCP server on first use and loads the skills. Nothing
else to configure.
Use it
Ask in plain language:
"Simulate an FM station at 100.3 MHz and build me a receiver."
"Here's a capture,
mystery.wav— survey what's on the air, then decode the strongest carrier."
I pick the right skill and drive the workflow end to end:
- survey-spectrum — what's present (PSD, signal detection, spectrogram)
- simulate-scene — register a simulated device from a description
- build-receiver — channelize, demodulate, validate, run, verify, save
- debug-no-signal — locate the fault when a receiver outputs nothing
- tx-experiment — closed-loop transmit-then-receive in simulation
- escape-hatch — drop to the Python library when no tool fits
Artifacts land under artifacts/ in the server's working directory; set
MARCONI_WORKSPACE to put them elsewhere.
License
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.