Most AI tools live in the browser. They’re great at writing, answering questions, searching the web. But the moment you need help with a desktop app, a mobile screen, or an embedded system — they’re blind.
SimSense changes that.
What is SimSense?
SimSense is a simulator platform that gives Claude a live window into any screen — not a screenshot, a real-time, interactive sim. Claude can see what’s happening, click buttons, type text, and respond to what it observes. All through a standard MCP connector.
You set up the connector once. After that, Claude can:
See your app’s current state
Interact with UI elements
Run automated test sequences
Guide users step-by-step through complex workflows
Why this matters
The gap between “AI can write code” and “AI can help you use software” is enormous. SimSense bridges it.
Imagine asking Claude to help debug a crash in your embedded Linux device — and Claude can actually look at the screen, navigate the menus, and reproduce the issue. Or asking it to run a regression test suite across five different device configurations, while you grab a coffee.
That’s what SimSense makes possible.
How it works
SimSense exposes your devices and simulators as MCP resources. Claude connects to SimSense via the MCP Server URL, discovers available sims, and can then:
Query the current screen state
Send input events (taps, keystrokes, swipes)
Read app state and logs
Compare expected vs. actual UI states
It’s all done through Claude’s tool-use — no custom prompt engineering required on your end.
Get started
Setting up the connector takes about two minutes:
Open Claude → Settings → Connectors
Add a custom connector named SimSense
Paste the MCP Server URL:
https://my.simsense.ai/mcpHit Connect and sign in
After that, just ask Claude to discover your sims and start working.
We’re just getting started. More device types, better state introspection, and team collaboration features are on the roadmap. If you’re building something interesting on top of SimSense, we’d love to hear about it
.