Drive a MetaClaw runtime from your browser.

The configurator is a web page. Your runtime, your solvers, and your design files stay on your machine. Open this page and it looks for a runtime listening on the loopback address, then connects.

Runtime detection idle

Each candidate answers a handshake request. The runtime is accepted only when it reports the expected web-UI runtime contract; a stale or foreign listener is refused, not adopted.

Runtime stays local

The control plane binds to 127.0.0.1 only. Your MetaHz launcher connects trusted pages automatically. Other hosted pages require the runtime's permission and a pairing code.

Solvers fail closed

Reduced-order routes run on request. Native routes such as HFSS or COMSOL stay gated until license, mesh-review, and convergence evidence are present.

LLM backend is yours to choose

Run planning locally through Ollama, or configure an online provider in LLM backend. Provider settings are saved on your connected runtime.

Download MetaClaw runtime

Install MetaClaw on this computer, connect the runtime, then choose your LLM and tool defaults in LLM backend.

MetaClaw for Windows

The installer downloads Node.js and sets up one private Python 3.11.7 environment with PyAEDT, WiTwin and PyTorch. It detects your NVIDIA GPU and driver and selects a compatible CUDA build, or uses CPU mode. An internet connection is required during setup. You do not need Conda, a source checkout, or an existing Python installation.

Download Windows installer (.exe)

Windows 10/11 x64 · 17.6 MiB · Self-signed installer — Windows does not automatically trust this publisher (MetaClaw). SHA-256: 4f30e2fa77c3935a5ca644cdf2f0f3110730a0142557ef84b88ddac1067f8635

Run the downloaded .exe and choose an installation folder.
Launch MetaClaw from its shortcut. Runtime Setup opens on an available local port.
Return here and click Connect. Allow local-network access if your browser asks.
A fresh installation connects automatically to metahz.ai. Existing connection settings are preserved on upgrades.

Choose the solver for your computer

Calculators and design editing run without a native solver. Runtime Setup reports which simulation tools are ready.

HFSS needs a supported AEDT installation and license. WiTwin antenna searches need the supported Linux NVIDIA environment or your configured solver connection. RIS coverage uses the separately configured Sionna service.

You can design, save and export antennas while setting up simulation.

Open Antenna Studio

Runtime setup

Runtime connection, version, environment and solver readiness.

Runtime connection

Connect to MetaClaw to view runtime information.

Select this computer or a configured runtime address in the connection bar above.

Runtime details and solver readiness

Runtime

Dispatch gates what would block a solver launch right now

-engines ready

-skill profiles

-solver profiles

-jobs this session

State round-trip last 40 polls, ms

LLM backend

Configure provider connections, choose the default LLM and set each tool's LLM and solver defaults.

Default LLM

Connect your runtime, configure a provider, then choose your default LLM.

Online models receive your prompts. Save API keys directly to your runtime using Configure provider.

Active selection

Choose Default LLM and any tool overrides, then Save defaults. Configure provider opens its connection settings on this page.

Tool defaults

Each tool uses its selected LLM and solver. Existing designs keep their own solver choice. Saving does not start a model request or a simulation.

Tool	LLM	Solver	Workspace
Unit Cell Designer	Connect to load saved LLM and solver defaults.
Metasurface Designer	Connect to load saved defaults.
Filter Designer	Connect to load saved defaults.
Antenna Studio	Connect to load saved defaults.
RF Calculator	Connect to load saved defaults.
Simulation planner	Connect to load saved defaults.

Connect to your runtime to configure tools.

Provider connections

Remote GPU · SSH

Save reusable SSH access to a remote GPU host. SSH does not require Ollama and does not change your local Ollama or solver settings. Enable remote Ollama only if you want that host to run LLM requests; it must already have Ollama and a model installed. Connecting does not install software or start a solver.

Connect MetaClaw to load saved SSH presets.

Saved SSH preset

Use this GPU preset selects the remote connection without changing LLM defaults or configuring a solver. Disconnecting closes SSH; reconnect explicitly to use it again.

Engines

Every engine the runtime knows about, with the readiness it reports for execution and for scientific claims.

Engine	Status	Execution	Scientific	Notes

METAHZ / SIMULATION WORKSPACE

Simulation Studio

Existing profile execution & job history

Use this when a reviewed skill already defines the model. A planning brief does not replace a native model or authorize a solve.

Dispatch

Engine

Skill profile

Arm solver dispatch

Arm first. Dispatch disarms automatically when the job is accepted.

Jobs idle

Design loop

Iterate a profile toward an objective through the runtime's design-iteration job. Results carry the evidence level of the route that produced them.

Loop request

Objective

Skill profile

Maximum iterations 8

Latest loop state

Evidence

What the runtime holds as its latest solver artifacts and reports. A file's presence is not proof; the labels say what was verified.

HFSS artifacts

Reports and exports

Bridges

Other ways to reach this runtime: a hosted configurator, the phone app, and chat channels.

Hosted configurator local page

A hosted copy of this page pairs with the runtime using this code.

------

Mobile app unknown

Pairing code for the phone

WhatsApp unknown

Console

The same command layer the CLI and chat bridges use. Try help or status.

Options

Configurator preferences. They live in this browser, not on the runtime.

Appearance

MetaHz shared theme · warm light

Connection

State poll interval