PicoScope MCP
PicoScope oscilloscopes connect to a computer over USB. An AI assistant can help write code for an experiment, but working with the instrument also requires a way to configure it and retrieve its samples. This project provides that connection through the Model Context Protocol (MCP).
The project uses a Python server between the assistant and PicoSDK, the library that controls the scope. The assistant can request operations such as selecting a channel, setting a trigger, and capturing a waveform. The server translates those requests into calls to the instrument library.
From a request to a waveform
A capture involves several decisions before any samples come back: which device to connect to, which channel to enable, what voltage range to use, and when to trigger. The server exposes these as separate operations so the assistant can configure the measurement explicitly. The block-capture path then returns the sampled waveform for inspection.
What is implemented
Device configuration and block acquisition have implementation paths, but they still need validation with the connected scope and its native SDK. Streaming is unfinished. Several analysis tools, including FFT, THD, and rise time, are placeholders; the frequency and amplitude tools currently return instructions rather than calculated measurements. The project is a prototype of the instrument connection, with more work needed before an assistant can rely on it for a complete measurement.