Rigol DS1202Z-E MCP
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Rigol DS1202Z-E MCP inspect the enabled channels and capture a screenshot of the current signal"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Rigol DS1202Z-E MCP
A Calligraphy Systems project.
Ask an AI for a measurement without making it manage SCPI commands, transfer formats or instrument sequencing. This Linux MCP server gives the Rigol DS1202Z-E a compact default interface and discoverable advanced operations.
AI client -- SSH / Wi-Fi -- Linux instrument host -- USB -- DS1202Z-EVersion 0.6.0. Tested and deployed on one DS1202Z-E running firmware 00.06.03.SP2, including physical 24M-point waveform and ESP32 protocol checks. Other models and firmware are not qualified. Independent software; no Rigol affiliation.
What it does
Inspect enabled channels with units, unavailable readings and acquisition context.
Configure channels/edge triggers and acquire screenshots or full waveform memory.
Capture single events, save supported setups and restore with readback checks.
Configure scope serial decoders, or capture/decode supported UART, I2C and SPI formats on the host and restore the prior core setup.
Analyze saved waveforms with FFT and retrieve bounded samples/events on demand.
Locate narrow spikes with compact min/max overviews, optional waveform images, and exact-sample follow-ups. Full raw captures stay on disk, outside model context.
Ten top-level tools stay fixed. Start with inspect_signal(); use
discover_operations(query="capture decode") for advanced tasks and request the
selected operation's schema before calling it. No arbitrary SCPI endpoint exists.
Related MCP server: siglent-sds-mcp
Quick start without hardware
Linux and Python 3.12+ are required. A virtual environment keeps system Python intact.
python3 -m venv .venv
.venv/bin/python -m pip install --require-hashes -r requirements.lock
.venv/bin/python -m pip install --no-deps -e .
.venv/bin/rigol-mcp --simulate --identify
.venv/bin/python -m pytest -qThe lock includes development/test dependencies. Build tools may need separate
installation. Start .venv/bin/rigol-mcp --simulate --allow-control from an MCP
client to exercise simulated operations; it speaks stdio, not a terminal prompt.
Simulation does not verify physical hardware or calibration.
Connect a real scope
Follow host and SSH setup. Use rigol-mcp --list-usb to
discover the actual VISA address, then pass --resource or set RIGOL_RESOURCE.
Controls require --allow-control. Physical probe ratio must match configuration.
Review wiring and operation prerequisites before changing acquisition/settings.
Documentation
Version 0.6.0 validation: 293 automated tests, 34 physical saved-waveform checks, and nine UART/SPI/I2C capture/decode/restore runs with all 158 events verified. This is functional evidence, not calibration or compliance certification. Private raw captures and machine configuration are excluded; the validation summary states tested conditions and gaps. Source catalog evidence paths refer to those historical private records; they are provenance labels, not downloadable public artifacts.
Completed artifacts persist on the host, subject to retention. Active operations are synchronous: continuing through SSH loss and durable background jobs are planned, not implemented. Ordinary setting changes do not automatically roll back.
Versions 0.5.1 and later reject legacy artifacts without metadata checksums: archive important evidence, complete any old-version restoration before upgrading, and take new snapshots after checking the actual state. No automatic migration is performed. The decode planner now caps legal dual-channel allocations at 600,000 points; impossible rounded window/rate combinations fail before instrument access.
Part of the Calligraphy Systems Hardware MCP Suite
This server is one piece of the Calligraphy Systems hardware MCP suite: open, locally run MCP servers that let an AI assistant work with real engineering tools and lab equipment. The suite also includes calligraphy-eda-mcp for PCB layout and design-rule verification. More at calligraphyeda.com.
License and dependencies
Project code is covered by MIT. Dependencies retain their own licenses and are installed separately. No vendor firmware, manuals, private flash backups, credentials or captured user data are distributed.
References: Rigol programming guide, MCP Python SDK, PyVISA-py.
Available Tools
10 toolsacquisitionA
Start, stop, or arm a single acquisition. Requires --allow-control.
Run/single may replace an existing captured waveform. Single returns before triggering.
Inputs: action (required). Returns: object
advanced_controlADestructive
Execute a discovered control or full-memory capture, requiring --allow-control.
Validates before writing; returns actual settings and any partial failure. Changes may invalidate a retained capture. Inspect readback before retrying. Settings changes do not generally roll back; decode.capture explicitly restores the previous core setup. Full-memory capture performs its documented transfer/acquisition cleanup.
Inputs: arguments (required), operation (required).
advanced_readARead-only
Execute a discovered read operation. Exact parameters come from discover_operations.
Screenshots return a saved artifact receipt; artifact.image explicitly retrieves the image. waveform.overview is a bounded saved-data envelope; waveform.image returns its optional PNG. Cannot change acquisition or channel/trigger settings.
Inputs: operation (required), arguments.
discover_operationsARead-only
Find advanced capabilities by task. Supply an operation ID to get its exact schema and effects.
No instrument access. Planned entries cannot execute; search results stay compact.
Inputs: query, operation. Returns: object
identifyARead-only
Read model, serial, firmware, and whether this is a simulator.
Inputs: no named parameters. Returns: object
inspect_signalARead-only
Start here: inspect all enabled channels, or one channel, in one call.
Returns acquisition freshness, frequency/amplitude readings with units, probe ratios, unavailable readings and concrete next actions. No setup, identify, or status call needed. Does not change acquisition or settings. Null readings are not proof of no signal.
Inputs: channel. Returns: object
measureBRead-only
Read a built-in measurement. Invalid/no-signal readings return null with valid=false.
Inputs: channel (required), measurement (required). Returns: object
set_timebaseA
Change horizontal scale in YT mode and read back the actual setting.
Requires --allow-control. Scope may round to a supported 1-2-5 scale.
Inputs: seconds_per_division (required). Returns: object
statusARead-only
Read acquisition status, timebase, channel scales, units, and probe ratios.
Inputs: no named parameters. Returns: object
waveformA
Read a compact screen-waveform summary from a STOPPED scope.
Set include_samples=true only when the individual time/amplitude arrays are needed. Changes waveform transfer source/format/range; does not stop or restart acquisition. Leaves transfer settings at BYTE/NORMAL. Not a full-memory capture.
Inputs: channel (required), include_samples. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
identify, acquisition, set_timebase, and the discover/advanced trio are cleanly separated. inspect_signal intentionally bundles readings that also appear in status and measure, so a few boundaries are fuzzy, but the descriptions make the intended entry point clear.
Names mix bare verbs (identify, measure), bare nouns (status, waveform, acquisition), verb_noun (inspect_signal, set_timebase), and adjective_noun (advanced_read, advanced_control). Multiword names are consistently snake_case and readable, but there is no single predictable convention.
10 tools is a well-scoped size for an oscilloscope MCP. The discover/advanced abstraction prevents a large flat command surface while keeping common workflows directly accessible.
Core read, inspect, measure, waveform, acquisition, and timebase workflows are covered, and advanced discovery/control extends the surface. Direct front-panel setup for vertical channel and trigger configuration is not exposed as dedicated tools, leaving notable but workable gaps.
Maintenance
Related MCP Connectors
Lets AI agents analyze API metrics, request logs, and traces, and manage apps and alerts.
Protocol-native energy infrastructure orchestration for AI data centers. Provides 46 MCP tools across 8 grid protocols (IEC-61850, DNP3, Modbus, OCPP, OpenADR, IEEE 2030.5, IEC 60870-5-104, ICCP) with 5 core API primitives: connect, dispatch, settle, comply, and intel. Enables AI agents to programmatically interact with substations, grid interfaces, and energy assets for real-time workload-grid coordination.
Capture, inspect, send and assert on webhooks and inbound email from your coding agent.
Anonymous webhook capture, inspection, waiting, and response configuration for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to directly control NI oscilloscopes (e.g., PXIe-5160/5164/5110) through the Model Context Protocol, including waveform acquisition, measurement, and configuration.MIT
- AlicenseCqualityDmaintenanceEnables AI tools to remotely control a SIGLENT SDS800X HD series oscilloscope, configure acquisition, capture waveforms, and analyze signal quality safely through high-level SCPI tools.204GPL 3.0
- AlicenseAqualityCmaintenanceEnables LLMs to control RIGOL oscilloscopes over SCPI via LAN or USB, including configuring channels/timebase/trigger, taking measurements, capturing waveforms and screenshots, decoding serial protocols, running FFT analysis, and controlling the built-in AFG.38MIT
- AlicenseAqualityCmaintenanceEnables MCP-compatible AI clients to operate a RIGOL MHO98 oscilloscope over USB, including configuration, waveform and measurement acquisition, FFT analysis, and protocol decoding.16MIT