Configuration and Troubleshooting

July 29, 2026 · View on GitHub

This page shows you how to configure the Simulink® Agentic Toolkit. For an overview of the Simulink Agentic Toolkit, see the README.

Requirements

MATLAB R2023a or later with Simulink
A supported AI coding agent
Simulink Test (optional) — required only for the model_test tool
Some skills require additional toolboxes (e.g., System Composer, Simscape, Stateflow). Check the requires-products field in each skill's manifest.yaml under skills-catalog/ for additional requirements.

Installing from Local Files (Offline Computer)

To install Simulink Agentic Toolkit in an offline or air-gapped environment, first download these artifacts on a computer with internet access and transfer them to the target machine or a shared location.

Artifact	Where to Get It
MCP server binary	Latest release — download the binary for your platform (e.g., `matlab-mcp-server-macos-arm64`, `matlab-mcp-server-windows-x64.exe`)
MCP server toolbox	Latest release — download `MATLABMCPServerToolbox.mltbx`
Simulink Agentic Toolkit	Clone or download from GitHub
MATLAB Agentic Toolkit	Clone or download from GitHub (only needed when installing with `Toolkit="matlab"`)

After you download these artifacts, run the setupAgenticToolkit command in your MATLAB command window with these name-value arguments.

Argument	Value
`MCPServerLocation`	Path to the MCP server binary download
`MCPToolboxLocation`	Path to the MATLAB toolbox (`.mltbx`) download
`MATLABAgenticToolkitLocation`	Path to the MATLAB Agentic Toolkit repository clone
`SimulinkAgenticToolkitLocation`	Path to the Simulink Agentic Toolkit repository clone

The installer downloads any artifact you do not provide locally. To prevent internet access and report an error if an artifact is unavailable, set Offline=true. For example, use this command to install Simulink Agentic Toolkit from local files.

setupAgenticToolkit("install", Offline=true,  ...
    MCPServerLocation="/shared/agentic-toolkits/bin/matlab-mcp-server-linux-x64", ...
    MCPToolboxLocation="/shared/agentic-toolkits/toolboxes/MATLABMCPServerToolbox.mltbx", ...
    SimulinkAgenticToolkitLocation="/shared/agentic-toolkits/simulink-agentic-toolkit")

Automated Setup Configuration

If you use automated setup using the 'setupAgenticToolkit' function, it writes two things: an MCP server configuration (so your agent can talk to MATLAB) and skill registrations (so your agent has Simulink expertise). The details vary by platform.

Platform	MCP Configuration	Skills Delivery	How To Update Toolkit
Claude Code	`~/.claude.json` (mcpServers)	`claude plugin` system	`setupAgenticToolkit("update")`
GitHub Copilot	VS Code user-profile `mcp.json`	`~/.agents/skills/` symlinks	`setupAgenticToolkit("update")`
OpenAI Codex	`~/.codex/config.toml`	`~/.agents/skills/` symlinks	`setupAgenticToolkit("update")`
Gemini CLI	`~/.gemini/settings.json`	`~/.agents/skills/` symlinks	`setupAgenticToolkit("update")`
Sourcegraph Amp	`~/.config/amp/settings.json`	`amp.skills.path` direct ref	`setupAgenticToolkit("update")`

How skill delivery works: Claude Code uses the native claude plugin system — setup registers a marketplace and installs plugins automatically. Other platforms discover skills from ~/.agents/skills/ via symbolic links that setup creates pointing to the installed toolkit. When you re-run install, the linked skills update automatically. If new skills are added, re-run configure to create the additional links.

Platform-Specific Notes

Claude Code — Setup writes MCP configuration to ~/.claude.json and registers skills via the claude plugin system (marketplace + plugin install). If the claude CLI is not on PATH, setup falls back to skill symlinks in ~/.claude/skills/.

GitHub Copilot — Setup writes global MCP config to the VS Code user-profile mcp.json (~/Library/Application Support/Code/User/mcp.json on macOS, ~/.config/Code/User/mcp.json on Linux, %APPDATA%\Code\User\mcp.json on Windows) and creates skill symlinks in ~/.agents/skills/. Reload VS Code after setup completes (Cmd/Ctrl + Shift + P, then "Developer: Reload Window").

OpenAI Codex — Setup writes ~/.codex/config.toml. Skills are installed as global symlinks in ~/.agents/skills/. After setup, you may want to tune two settings in the [mcp_servers.matlab] section of ~/.codex/config.toml:

tool_timeout_sec = 600 — increases the tool timeout from the default (which is too short for many MATLAB operations like test suites and simulations). Increase further for very long-running tasks.
env_vars = ['WINDIR'] — Windows only. Required for Simulink to work, since Codex strips environment variables from MCP server subprocesses by default.

Gemini CLI — Setup writes global config to ~/.gemini/settings.json and creates skill symlinks in ~/.agents/skills/. Start a new Gemini session after setup.

Sourcegraph Amp — Setup writes to ~/.config/amp/settings.json using the amp. prefix for all keys. Skills load directly from the toolkit via amp.skills.path (no symlinks needed). If you have amp.mcpPermissions rules that block MCP servers, setup will detect this and ask before making changes.

Manual Setup Configuration

If you prefer to manage your own MCP server installation and agent configuration, you can set up the toolkit manually following the instructions in the MATLAB MCP Server repository to install the MCP server binary and configure it with your coding agent. For an overview of manual set up, see the README.

Update Simulink Agentic Toolkit

Run the update action in MATLAB to download the latest toolkit and MCP server:

setupAgenticToolkit("update")

After updating:

Re-run satk_initialize in MATLAB to pick up any tool changes
Restart your agent session to load updated skills

Other Setup Actions

The setupAgenticToolkit function supports several actions:

Action	Command	Description
Install	`setupAgenticToolkit("install")`	Download MCP server and toolkit files, then configure
Configure	`setupAgenticToolkit("configure")`	Set up an agent with MCP and skills
Update	`setupAgenticToolkit("update")`	Download latest MCP server and toolkit files
Uninstall	`setupAgenticToolkit("uninstall")`	Remove installed toolkits and agent configurations
Status	`setupAgenticToolkit("status")`	Show current installation and configuration status

All actions support Prompt=false for non-interactive use.

setupAgenticToolkit("install", Toolkit=["matlab", "simulink"], Prompt=false)
setupAgenticToolkit("configure", Agents="claude-code", Scope="global", Prompt=false)

If your organization uses a CLI wrapper, pass AgentCLI="claude-code=/path/to/wrapper" during configure.

Custom Agent CLI Commands

If your organization uses a wrapper or alias for agent CLI binaries, use the AgentCLI parameter:

setupAgenticToolkit("configure", AgentCLI="claude-code=/usr/local/bin/my-claude-wrapper")

The format is "agent-id=command". This override is saved to config.json and automatically used for all subsequent actions (configure, uninstall). You only need to specify it once.

Note: Currently only Claude Code uses a CLI during setup (for plugin registration via claude plugin). All other agents are configured via file writes and symlinks, so they do not need AgentCLI. If the Claude CLI is not found, setup falls back to symlinks automatically.

Remove Agent Configurations

To remove agent configurations without uninstalling toolkits, run:

setupAgenticToolkit("uninstall")

Then select Agent configurations only from the interactive prompt. This removes MCP config entries and skill registrations while keeping installed toolkits and the MCP server intact. Useful when switching agents or cleaning up stale configurations.

Disable Data Collection

MATLAB MCP Server collects fully anonymized information about your usage of the server and sends it to MathWorks. This data collection helps MathWorks improve products and is on by default. To opt out of data collection, configure the toolkit with the DisableTelemetry option set to true by running this command in MATLAB:

setupAgenticToolkit("configure", DisableTelemetry=true)

This command opts every configured agent out of data collection. This setting is retained when you update to a new version the toolkit with setupAgenticToolkit("update"). If you reconfigure the toolkit for your agent(s) by running setupAgenticToolkit("configure"), include DisableTelemetry=true again to keep data collection disabled.

Troubleshooting

Problem	Likely Cause	Fix
Add-On Manager fails when opening mltbx (`ERR_CERT_AUTHORITY_INVALID` or "Unable to open the requested feature")	CEF/display issue — often caused by corporate proxies, antivirus software, or headless environments	Install programmatically instead: `matlab.addons.toolbox.installToolbox("agenticToolkitInstaller.mltbx")`
Agent doesn't list Simulink skills	Skills not registered	Re-run `setupAgenticToolkit("configure")`
MCP tools fail with "Undefined function"	`satk_initialize` not run in current MATLAB session	Run `satk_initialize` in MATLAB
MCP server can't connect to MATLAB	Connector not running or stale connection	Add `--log-folder` and `--log-level` arguments to your MCP server configuration (see MATLAB MCP Server arguments), then run `satk_initialize` again (it calls `shareMATLABSession` automatically). Check the generated logs.
macOS blocks the MCP server binary	Gatekeeper quarantine	Right-click → Open, or run: `xattr -d com.apple.quarantine ~/.matlab/agentic-toolkits/bin/matlab-mcp-server`
"rmiml.selectionLinkHelper" error	Path corruption from other toolboxes	Run `restoredefaultpath` in MATLAB, then re-run `satk_initialize`
`model_test` fails or is unavailable	Simulink Test not installed	Install Simulink Test, or use the other 7 tools which work without it
Codex tool calls time out	Default tool timeout too short for MATLAB	Add `tool_timeout_sec = 600` (or higher) to `[mcp_servers.matlab]` in `~/.codex/config.toml`
Simulink fails in Codex on Windows	Missing `WINDIR` environment variable	Add `env_vars = ['WINDIR']` to `[mcp_servers.matlab]` in `~/.codex/config.toml`

Report Bugs

If you encounter a bug, use the filing-bug-reports skill to generate a report before opening a GitHub issue. Ask your agent:

File a bug report for this issue

The skill automatically captures environment details, reproduction steps, and error output. Run the skill in the same session where the bug occurred, because it uses conversation context to reconstruct what happened. Then open a bug report and paste the generated report.