A sandbox of experimental Agent Skills for MATLAB and Simulink work, covering MATLAB tutoring, debugging coaching, hands-on exercises, and course AI-use policies. Skills may change without notice.
Claude Code
$ npx -y skills add matlab/agent-skills-playground --agent claude-code
Codex
$ npx skills add matlab/agent-skills-playground -a codex
Cursor
$ npx skills add matlab/agent-skills-playground -a cursor
Gemini CLI
$ npx skills add matlab/agent-skills-playground -a gemini-cli
GitHub Copilot
$ npx skills add matlab/agent-skills-playground -a github-copilot
OpenCode
$ npx skills add matlab/agent-skills-playground -a opencode
Windsurf
$ npx skills add matlab/agent-skills-playground -a windsurf
A sandbox for prototyping and demonstrating Agent Skills for MATLAB and Simulink work. Skills here are experimental. They may be incomplete, change without notice, or migrate to an official toolkit over time.
For official skills, use Agentic Toolkits for MATLAB and Simulink:
- MATLAB Agentic Toolkit: Gives your AI agent the knowledge and context to work efficiently with MATLAB and its toolboxes
- Simulink Agentic Toolkit: Give your AI coding agent the ability to read, build, edit, and test Simulink® models using Model-Based Design best practices.
| Demo | What it shows |
|---|---|
| engineering-an-agent-skill | A test-first process for authoring your own Agent Skills, with a meta-skill that walks an agent through the five stages. Accompanies the blog post. |
| mbse-with-agentic-ai | Agent-driven Model-Based Systems Engineering workflow in MATLAB and Simulink (RFLP methodology). Accompanies the blog post. |
| requirements-based-verification | Generate Simulink Test Sequence harnesses from reviewed EARS requirements. Includes the skill, a ModeLogic example, and MATLAB tests. |
| embedded-ai-deployment | Agent-driven deployment of AI models to embedded hardware with MATLAB, Simulink, and Embedded Coder. Routes between native models compressed for lean targets (Cortex-M, DSP) and direct C/C++ generation from PyTorch and LiteRT models for high-performance hardware. Requires MATLAB R2026a. |
| lorenz-uihtml-app | An interactive Lorenz attractor, built by an agent using the matlab-uihtml-app-builder and matlab-uihtml-design skills. An HTML control panel drives an ode45 integration, with animated tracing, live speed control, mid-run stop, and PNG export. |
| boids-uihtml-app | An interactive Boids flocking simulator for learning emergence, built with the same two uihtml skills in the Warm Dark style. Live rule-weight sliders, behavior presets, a flock-order readout, per-boid inspection, and predator strikes. The README teaches the model through guided experiments. |
| assessment-generation-for-matlab-grader | Agent-driven generation of MATLAB Grader assessment items from a learning objective: descriptions, reference solutions, learner templates, randomized tests, and optional QTI 3 interchange files. Includes an instructor setup-guide skill for course rollouts. |
| ai-tutoring | A coordinated set of 11 skills that turn an agent into a course-aware MATLAB tutor: concept coaching, evidence-based debugging, practice activities, AI-use policies and guardrails, and session reports. Includes an instructor demo kit and example reports. |
| course-generation | Agent-driven generation of complete MATLAB and Simulink courses in IMS Common Cartridge format, following the MATLAB Course Designer model. An instructor interview feeds the IDStack instructional-design pipeline; MATLAB Exercises are validated against a live MATLAB session and Simulink Exercises get real starter and solution models. |
Standalone skills under skills/, each installable on its own.
Demos bundle their own copies separately.
| Skill | What it does |
|---|---|
| simulink-test-sequence-from-ears | Generates Simulink Test Sequence harnesses from EARS requirements and model signals, buses, and enumerations. Uses a YAML mapping review before generation and compile-checks the harnesses. |
| matlab-performance-optimizer | Speeds up slow MATLAB code: vectorization in place of loops, preallocation, appropriate data types, sparse matrices, and in-place operations. Uses the profiler and timeit to find the real bottleneck before changing anything. |
| matlab-uihtml-app-builder | Builds interactive MATLAB apps with an HTML and JavaScript front end over a MATLAB computational backend, bridged by the uihtml component. Covers app structure, event wiring, and the communication patterns in both directions. |
| matlab-uihtml-design | Generates production-grade HTML, CSS, and JavaScript control panels for uihtml. Ships eight built-in styles with an interactive gallery for picking one, a base template per style, and a full design specification for each. Also takes a custom aesthetic described in prose. |
These skills started here and now ship in an official toolkit. Their folders here contain only a pointer to the official version. If you installed one from this repository, replace it with the official skill.
| Former skill | Official skill |
|---|---|
| matlab-symbolic-math | matlab-use-symbolic-math in the MATLAB Agentic Toolkit, Math and Optimization group |
Skills are modular, portable capabilities that work across coding agents:
See the Agent Skills specification for the file format.
git clone https://github.com/matlab/agent-skills-playground.git
cp -r agent-skills-playground/skills/<skill-name> ~/.claude/skills/
For Claude.ai (web) and Claude Desktop, zip a skill's directory (with SKILL.md at the root) and upload it via Settings, then Capabilities, then Upload Skill.
For other agents, follow the host's instructions for installing user-defined skills.
git clone https://github.com/matlab/agent-skills-playground.git
cd agent-skills-playground/demos/<demo-name>
Each demo has a tutorial README and supporting files. Demos that introduce skills bundle them under skills/; app demos link to the standalone skills used to build them. Follow the demo's README to install its skills and prepare the example before opening it in your agent. A folder named skills/ is not automatically discovered by every agent.
The license is available in the LICENSE.md file in this GitHub repository.
Copyright 2026 The MathWorks, Inc.
Supported coding agents: Agent Skills work with Claude Code, Claude.ai, Cursor, VS Code, GitHub Copilot, Gemini CLI, OpenAI Codex, Amp, Goose, Roo Code, and many more. Feature availability varies by plan and platform.
Stars on GitHub
View @matlab’s profileHelps MATLAB and Simulink users with tutoring, debugging and course AI-use policies. Covers coaching through errors and failed tests, creating hands-on exercises, and writing an AI-use policy file for a course.
@matlabA development methodology for coding agents that makes the agent brainstorm before building, follow plans, run parallel work and finish branches in a disciplined way. You get fewer skipped steps and more consistent results from your coding assistant.
@obraHelps you take a change from a rough idea or bug report through to a tested, reviewed result, with each step handled by a focused skill. Its skills cover pinning down domain terms, designing module interfaces, diagnosing hard bugs, and reviewing changes against your repo's coding standards.
@mattpocockGives your coding agent a planning, testing and review routine to follow on every task. It plans before building, checks changes with tests, has a fresh reviewer look for regressions, and saves repeated wins as reusable skills.
@affaan-mDeveloper tools for working on the React codebase itself. Provides skills for extracting error messages, managing and debugging feature flags across release channels, running Flow type checks, and fixing lint and formatting issues before committing.
@facebookInstall it with the command on its page. It includes matlab-apply-assignment-guardrails, matlab-coach-debugging, matlab-coach-programming. Its skills do not fire on their own yet. Free and open source.
Is this plugin yours?
Claim it with GitHubSubmit a pluginPromote it