Tidy3D for VS Code

Build, visualize, and iterate on Tidy3D FDTD simulations right in Visual Studio Code. The extension discovers simulations in Python files and notebooks, opens a 3D viewer next to your code, and (after setup) lets Copilot Chat use the bundled Tidy3D MCP server for contextual docs and examples.

Highlights

• Editor‑native 3D viewer – open simulations alongside your code and see updates as you tweak parameters.
• One‑step setup – register the bundled MCP server, import FlexAgent rules, and connect the latest docs.
• Shareable insights – capture screenshots, toggle structures, or let Copilot control the viewer.

Prerequisites

• Visual Studio Code 1.99.3 or newer.
• Python extension enabled for notebooks and Python files.
• Flexcompute Tidy3D account with an API key: tidy3d.simulation.cloud/account?tab=apikey.
• You don’t need a local Python setup to view simulations. If you plan to execute notebook cells, enable the Python extension and select an interpreter as usual.
• Internet access for the first‑time download of viewer assets and runtimes.

Install the extension

1. Open Visual Studio Code and go to the Extensions view (View → Extensions).
2. Search for Tidy3D by Flexcompute and select Install. Listing: Visual Studio Marketplace.
3. Confirm the Python extension is enabled. Restart VS Code if prompted.

Connect your Tidy3D account

1. Open the workspace that holds your notebooks or Python scripts.
2. Run Cmd/Ctrl+Shift+P → Tidy3D: Run Tidy3D Workspace Setup (or accept the prompt).
3. When asked, paste your API key. The key is stored in your VS Code settings and can be rotated later from the Tidy3D settings panel.
4. Setup registers the bundled MCP server, writes FlexAgent rules under .github/instructions/, and validates connectivity before enabling the viewer.

Optional: prepare your Python environment (for notebooks)

1. Create or select a virtual environment for your notebooks (venv, conda, poetry, rye, hatch, etc.).
2. Install the basics: tidy3d, jupyter (or jupyterlab), numpy.
3. Example with uv:
   • Install uv from the official guide.
   • Define dependencies in pyproject.toml.
   • Run uv sync.

Quickstart notebook

1. Download the official Quickstart: StartHere.ipynb.
2. Open it in VS Code. If you plan to execute notebook cells, select your Python environment; the viewer itself does not require one.
3. Add or open a cell that defines a simulation (e.g., td.Simulation(...)). Detection is automatic; you do not need to execute the cell for the viewer to pick it up.
4. Open the viewer via any of:
   • Status bar “Tidy3D: N simulations” entry (click)
   • Command Palette: Tidy3D: Open Tidy3D
   • Python editor context menu (for .py files)
5. Pick a simulation in the quick pick and explore fields. Edits to code or notebook cells update the viewer automatically.

Work with Copilot

• Ask Copilot to “Validate my Tidy3D simulation” to open the viewer, check mesh quality, and flag configuration issues.
• Explore field slices, materials, or sources while Copilot steers the viewer for you.
• Request design alternatives; Copilot can edit notebooks, rerun simulations, and keep the viewer synced.

Enable Copilot Chat instructions

After you run “Tidy3D: Run Tidy3D Workspace Setup”, the extension drops instruction files under .github/instructions/.

• In VS Code Settings, enable or verify: github.copilot.chat.codeGeneration.useInstructionFiles.
• This lets Copilot Chat automatically use the workspace instructions in chat. You can confirm by sending a chat message and expanding “References” to see the loaded instruction files.

Example prompts

• Learn the method – “Explain how FDTD in Tidy3D updates the fields in this notebook.”
• API look‑up – “Find the Tidy3D API for a GaussianBeam source and add it.”
• Troubleshoot – “The run failed with a boundary condition error. Diagnose and fix it.”
• Optimize – “Sweep the waveguide width and plot transmission for each run.”

Commands

• Tidy3D: Open Tidy3D (tidy3d.openViewer) – Command Palette, Python editor context menu (.py), or status bar.
• Tidy3D: Run Tidy3D Workspace Setup (tidy3d.installWorkspaceSetup) – Command Palette.

Settings

• tidy3d.apiKey (default: empty) – API key for MCP authentication (workspace‑scoped).
• tidy3d.pythonVersion (default: 3.12) – Python version used by the built‑in runtime.
• tidy3d.viewer.remoteUrl (default: https://tidy3d.simulation.cloud/simulation-viewer) – Remote viewer UI.
• tidy3d.viewer.tidyVersion (default: empty) – Override Tidy3D Python package version used by the viewer.
• tidy3d.screenshotsDir (default: reports/.tidy3d-screenshots) – Folder for viewer screenshots.
• tidy3d.viewerBridge.port (default: 0) – 0 chooses a free port; set a fixed port to allow external clients.
• tidy3d.viewerBridge.bindAddress (default: 127.0.0.1) – Interface the local viewer bridge binds to.
• tidy3d.mcp.remoteUrl (advanced) – Base URL for the Flexcompute MCP service.
• tidy3d.mcp.runCommand (advanced) – Custom command to launch the MCP server; standard args are appended automatically.
• tidy3d.mcp.enableViewer (default: true) – Enables viewer tooling when launching the MCP server.

Privacy and network usage

• First run downloads the Python runtime and viewer bundles; later sessions reuse cached artifacts.
• Simulation data stays local to your VS Code session. Network access is used for extension updates, static viewer assets, and the remote viewer UI when used.

Troubleshooting

• No simulations detected – ensure a variable receives td.Simulation(...) (or another supported solver type) in the active file.
• Workspace setup missing – confirm the Python extension is active and the file uses the python language mode.
• Slow first run – wait for initial downloads; you’ll see “Tidy3D Kernel: ready” in the status bar when ready.
• Viewer cannot connect – verify your API key and outbound access to tidy3d.simulation.cloud.

License

Proprietary – Flexcompute Inc.