Language: English | 简体中文
Simulink Automation Suite
[!IMPORTANT]
This project is paused (since June 2026). MathWorks has released an official, actively-maintained agentic toolkit for Simulink — for production use, please switch to the official Simulink Agentic Toolkit.
The differentiators explored here (the guarded write-safety model: dry-run / rollback / read-back verification; and post-simulation performance analysis), together with a new motor-control-focused direction, will continue in a separate, dedicated project. This repository remains available as-is for reference.
Simulink Automation Suite is a Claude Code plugin for Simulink automation workflows through MATLAB Engine for Python.
- Canonical plugin name:
simulink-automation-suite
- Read Analysis — the
simulink-analyzer agent autonomously explores model topology, traces connections, audits parameters, and returns structured findings without polluting conversation context.
- Post-Simulation Analysis — the
sim-analyst agent extracts signals from sl_sim_result, evaluates dynamic performance (rise time, overshoot, settling time), and returns structured conclusions without exposing raw waveform data to the main conversation.
- Write Automation — the
simulink-automation skill guides safe parameter modification with dry-run preview, precondition guards, and rollback support.
- Runtime Python module path:
simulink_cli (unified CLI entrypoint)
Positioning
Simulink Automation Suite is built to make Simulink analysis agent-native in Claude Code:
- It exposes Simulink context as deterministic, machine-readable tool outputs.
- It lets AI reason on real model topology/parameters instead of screenshots.
- It keeps workflows real-time and token-efficient with clipping/projection controls.
In short: the plugin helps AI understand first, then assist.
Why This Plugin Exists
Common AI+Simulink workflows are often one of these:
- Screenshot-based discussion: fast but shallow, visual-only understanding.
- Export-and-parse flow: richer context but heavy, delayed, and token-expensive.
This plugin provides a third path: direct, structured, runtime model analysis for agents.
How It Works
- Claude Code invokes the
simulink-automation skill for write/meta tasks, or dispatches the simulink-analyzer agent for read analysis.
- The skill resolves MATLAB session context (
session list/use/current/clear) with exact-name matching, using either an explicit --session or a previously selected active session.
- It executes one of the 18 available actions:
schema, list_opened, scan, connections, inspect, find, highlight, set_param, model_new, model_open, model_save, model_close, model_update, block_add, line_add, line_delete, block_delete, simulate, or session.
- Results are returned as a single machine-readable JSON payload on
stdout; warnings never spill raw text into stdout, and stderr is reserved for maintainer-facing diagnostics.
- Failures use stable error codes for reliable agent recovery.
- Write operations (
set_param) use dry-run preview (default), machine-executable apply_payload, rollback payloads, guarded execute via expected_current_value, and read-back verification.
Prerequisites
Before using session-bound actions (list_opened, scan, connections, inspect, find, highlight):
- Install and activate MATLAB on your machine.
- Install MATLAB Engine for Python in the same Python interpreter that runs this plugin.
- In MATLAB Command Window, run:
matlab.engine.shareEngine
Troubleshooting:
engine_unavailable: MATLAB Engine for Python is unavailable in the active Python environment. Fix interpreter/environment installation.no_session: MATLAB Engine is available, but no shared MATLAB session is visible. Run matlab.engine.shareEngine in MATLAB, then retry.
Quick Start
1. Add the marketplace source
/plugin marketplace add Mistakey/simulink-automation-suite
2. Install the plugin from marketplace
/plugin install simulink-automation-suite@simulink-automation-marketplace
3. Invoke the namespaced skill
/simulink-automation-suite:simulink-automation Scan gmp_pmsm_sensored_sil_mdl recursively and focus on controller subsystems.
4. Verify plugin registration (optional)
/plugin list simulink-automation-suite@simulink-automation-marketplace
Scenario Examples
