Radar Designer MCP Control

Launch and control the MATLAB Radar Designer app programmatically via MCP. Use this skill when a user asks about radar design parameters, wants to configure a radar system, compare radar types, analyze radar performance, or visualize radar metrics interactively in the Radar Designer app.

On Entry

When this skill is first invoked without a specific user request, present the following example prompts to inspire the user:

1. Design a tracking radar at 5 GHz with 2 MW peak power at 1500 km range for a 1 sq meter target
2. Compare airport radar performance at 2.8 GHz vs 5.6 GHz — which can achieve 150 km range for a 10 sq meter target?
3. Set up a weather radar and show how heavy rain (16 mm/hr) degrades detection range
4. Configure an automotive radar at 77 GHz with electronic scanning for 100 meters range
5. Add a 200 km max range requirement to my current radar design and check if it meets the objective

Then wait for the user to type their own radar design question.

IMPORTANT

• Never generate MATLAB scripts for the user that call the wrapper functions directly. These are for the agent's internal use only. When the user asks for a script, use only the built-in export commands via radarDesignerExport.
• When greeting the user, simply ask about their radar design goals (type, frequency, power, requirements) without referencing the underlying mechanism.

When to Use

• User asks to design, configure, or analyze a radar system
• User wants to launch or open the Radar Designer app
• User asks to change radar parameters (frequency, power, antenna, etc.)
• User wants to compare different radar types (tracking, airport, airborne, etc.)
• User asks about radar performance metrics (max range, SNR, detection probability, etc.)
• User wants to load or save a radar design session (.mat file)
• User asks about target or environment configuration for radar analysis
• User wants to set requirements/objectives for a radar design

When NOT to Use

• User asks about general MATLAB programming unrelated to radar design
• User wants to use Radar Toolbox functions directly (without the Radar Designer app)
• User asks about Simulink radar models or Phased Array System Toolbox without the app
• User wants to create radar waveforms or signals outside the app context
• User asks about radar theory or equations without wanting to use the app

Prerequisites

• MATLAB MCP server must be running and connected
• The mcp__matlab__evaluate_matlab_code tool must be available
• The Radar Toolbox must be installed in the MATLAB instance
• The skill's scripts/ directory must be on the MATLAB path (set via project_path)

Code Reference

Consult code-reference.md for detailed code patterns — including multi-radar comparison, session management, export, and range auto-tuning examples. The patterns below are summaries; code-reference.md is authoritative.

Wrapper Scripts

All Radar Designer operations go through 5 p-coded wrapper scripts in scripts/. Always set project_path to the skill's root directory when calling mcp__matlab__evaluate_matlab_code so the scripts are on the MATLAB path.

radarDesignerSession — App lifecycle and sessions

Templates: 'AirborneRadarSpec', 'AirportRadarSpec', 'AutomotiveRadarSpec', 'TrackingRadarSpec', 'WeatherRadarSpec'

radarDesignerParam — Get/set parameters and requirements

specType: 'Radar', 'Target', 'Environment'

radarDesignerResults — Read results and auto-tune

Results table columns: Metric, Units, Threshold, Objective, Result_<radarName>, Status_<radarName> (PASS/WARN/FAIL)

Auto-tune result fields: requestedRange_km, achievedRange_km, ratio, converged

radarDesignerMultiRadar — Multiple radar management

Add templates: 'AirborneRadar', 'AirportRadar', 'AutomotiveRadar', 'TrackingRadar', 'WeatherRadar'

radarDesignerExport — Built-in export

Workflow

Step 1: Launch the App

h = radarDesignerSession('launch');

Always check if h already exists:

if ~exist('h','var') || ~isvalid(h)
    h = radarDesignerSession('launch');
end

Step 2: Select a Radar Template

radarDesignerSession('startNew', h, 'TrackingRadarSpec');

Step 3: Set Parameters

radarDesignerParam(h, 'set', 'Radar', 'Frequency', 5e9);     % 5 GHz
radarDesignerParam(h, 'set', 'Radar', 'PeakPower', 2e6);     % 2 MW
radarDesignerParam(h, 'set', 'Target', 'RCS', 1);             % 1 m²
radarDesignerParam(h, 'set', 'Environment', 'RainRate', 4);   % 4 mm/hr

All values in SI units: frequency in Hz, power in W, range in m, etc.

Step 3a: Map User Objectives to Requirements

When the user states a performance goal, set it as the Threshold on the matching requirement.

Set both Threshold and Objective to the same value unless the user distinguishes a minimum acceptable (Threshold) from a desired goal (Objective).

Step 3b: Range Auto-Tuning (MANDATORY when user specifies a range)

If the user specifies a target range, you MUST run auto-tune BEFORE reading/reporting results:

result = radarDesignerResults(h, 'autoTune', 300e3);  % 300 km target

The auto-tune adjusts peak power (and gain if needed) so the achieved range is within ±15% of the user's request. It also sets the MaxRange requirement threshold to the user's requested range.

Skip this step ONLY if the user did not mention a specific range target.

Step 4: Read Analysis Results

T = radarDesignerResults(h, 'read');

The returned table has 16 rows (one per metric) with columns: Metric, Units, Threshold, Objective, and per-radar Result_<name> and Status_<name> columns.

Status values: PASS (meets objective), WARN (between threshold and objective), FAIL (does not meet threshold).

Step 5: Multi-Radar, Sessions, Export

See code-reference.md for complete patterns.

Complete Property Reference

Radar Properties (Settable)

Waveform

Noise

Antenna

Scanning

Detection

Track Confirmation

Signal Processing

Target Properties

Environment Properties

Requirement Specifications

Each requirement has: Objective, Threshold.

Presenting Results

Always present results to the user in the following structured format:

1. Summary Table

Use Unicode box-drawing characters to render a clean table.

Single radar:

┌───────────────────────┬────────────┬────────┐
│        Metric         │   Value    │ Status │
├───────────────────────┼────────────┼────────┤
│ Max Range             │ 315.1 km   │ PASS   │
├───────────────────────┼────────────┼────────┤
│ Range Resolution      │ 7.5 m      │ PASS   │
├───────────────────────┼────────────┼────────┤
│ First Blind Speed     │ 53.5 m/s   │ WARN   │
└───────────────────────┴────────────┴────────┘

Multi-radar comparison:

┌───────────────────────┬────────────┬────────────┬─────────┐
│        Metric         │  Radar A   │  Radar B   │ Winner  │
├───────────────────────┼────────────┼────────────┼─────────┤
│ Max Range             │ 315.1 km   │ 222.6 km   │ Radar A │
├───────────────────────┼────────────┼────────────┼─────────┤
│ First Blind Speed     │ 53.5 m/s   │ 26.8 m/s   │ Radar A │
└───────────────────────┴────────────┴────────────┴─────────┘

Table rules:

• Show only metrics that are relevant to the user's question or that differ meaningfully between radars (not all 16 by default)
• If the user asks for a "full report", show all 16 metrics
• Include display units in the value cells (km, m/s, dBW, etc.)
• Use the Status column values directly from the results table (PASS/WARN/FAIL)

2. Key Takeaways

After the table, provide 2-4 bullet points covering:

• The most significant performance tradeoffs
• Any surprising results or requirement failures
• What factor is limiting performance (e.g., rain loss, duty cycle, antenna gain)
• Impact of the user's parameter change (if they modified something)

3. Verdict

End with a 1-2 sentence summary or recommendation.

Conventions

1. UI Auto-Updates: All parameter changes fire events that update the web UI automatically — no manual refresh needed.

2. Handle Persistence: The variable h must persist in the MATLAB base workspace between calls. Each evaluate_matlab_code call shares the same workspace, so h remains available.

3. Startup Timing: The app takes 5-10 seconds to initialize. radarDesignerSession('launch') handles the wait automatically.

4. Always Use project_path: When calling mcp__matlab__evaluate_matlab_code, set project_path to the skill's root directory so the wrapper scripts in scripts/ are on the MATLAB path.

5. Property Names Are Case-Sensitive: Use exact names from the property reference tables above.

6. Range Auto-Tuning Is Mandatory: When the user specifies a target range, ALWAYS run radarDesignerResults(h, 'autoTune', range_m) BEFORE reading results. Never report a design that overshoots or undershoots the user's requested range by more than 30%.

7. Always Report Pass/Fail: When presenting results, always include the Status (PASS/WARN/FAIL) for each metric and explicitly call out which requirements are not met.

8. SI Units: All parameter values must be in SI units — frequency in Hz (not GHz), power in W (not MW), range in m (not km).

9. Always Consult code-reference.md: Before writing code for multi-radar comparison, session management, export, or before/after analysis, load and follow the patterns in code-reference.md.

Copyright 2026 The MathWorks, Inc.